feat: sub2api T05/T08/T09/T13/T14 + bump to 3.0.0-rc.7

T01 (P1): requested_model column in call_logs
- Migration 009_requested_model.sql: ALTER TABLE call_logs ADD COLUMN requested_model
- callLogs.ts: INSERT + SELECT updated to include requestedModel field

T02 (P1): Strip empty text blocks from nested tool_result.content
- New stripEmptyTextBlocks() recursive helper in openai-to-claude.ts
- Applied on tool_result content before forwarding to Anthropic
- Prevents 400 'text content blocks must be non-empty' errors

T03 (P1): Parse x-codex-5h-*/x-codex-7d-* headers for precise quota reset
- parseCodexQuotaHeaders() in codex.ts extracts usage/limit/resetAt
- getCodexResetTime() returns furthest-out reset timestamp for safe unblocking

T04 (P1): X-Session-Id header for external sticky routing
- extractExternalSessionId() in sessionManager.ts reads x-session-id,
  x-omniroute-session, session-id headers with 'ext:' prefix to avoid collisions

T06 (P2): account_deactivated permanent expired status on 401
- ACCOUNT_DEACTIVATED_SIGNALS constant + isAccountDeactivated() in accountFallback.ts
- Returns 1-year cooldown (effectively permanent) to prevent retrying dead accounts

T07 (P2): X-Forwarded-For IP validation
- New src/lib/ipUtils.ts with extractClientIp() and getClientIpFromRequest()
- Skips 'unknown'/non-IP entries in X-Forwarded-For chain

T10 (P2): credits_exhausted distinct account status
- CREDITS_EXHAUSTED_SIGNALS + isCreditsExhausted() in accountFallback.ts
- Returns 1h cooldown with creditsExhausted flag, distinct from rate_limit 429

T11 (P1): max reasoning_effort -> budget_tokens: 131072
- EFFORT_BUDGETS and THINKING_LEVEL_MAP updated with max: 131072, xhigh: 131072
- Reverse mapping now returns 'max' for full-budget responses
- Unit test updated to expect 'max' (was 'high')

T12 (P3): Model pricing updates
- MiniMax M2.7 / MiniMax-M2.7 / minimax-m2.7-highspeed pricing added

T15 (P1): Array content normalization for system/tool messages
- normalizeContentToString() helper exported from openai-to-claude.ts
- System messages with array content now correctly collapsed to string

.agents/workflows/generate-release.md

@@ -4,16 +4,36 @@ description: Create a new release, bump version up to 1.x.10 threshold, update c
 
 # Generate Release Workflow
 
-Bump version, finalize CHANGELOG, commit, tag, push, publish to npm, and create GitHub release.
+Bump version, finalize CHANGELOG, commit, open a **PR to main** and wait for user confirmation before tagging, publishing, and deploying.
 
 > **VERSION RULE: Always use PATCH bumps (2.x.y → 2.x.y+1)**
 > NEVER use `npm version minor` or `npm version major`.
 > Always use: `npm version patch --no-git-tag-version`
 > The threshold rule: when `y` reaches 10, bump to `2.(x+1).0` — e.g. `2.1.10` → `2.2.0`.
 
-## Steps
+---
 
-### 1. Determine new version
+## ⚠️ Two-Phase Flow
+
+```
+Phase 1 (automated): bump → docs → i18n → commit → push → open PR
+  ↕  🛑 STOP: Notify user, wait for PR confirmation
+Phase 2 (post-merge): tag → publish → GitHub release → Docker → deploy
+```
+
+**NEVER push directly to main or create tags before the user confirms the PR.**
+
+---
+
+## Phase 1: Pre-Merge
+
+### 1. Create release branch
+
+```bash
+git checkout -b release/v2.x.y
+```
+
+### 2. Determine new version
 
 Check current version in `package.json` and increment the **patch** number only:
 
@@ -27,11 +47,6 @@ Version format: `2.x.y` — examples:
 - `2.1.9` → `2.1.10` (patch)
 - `2.1.10` → `2.2.0` (minor threshold — do manually with `sed`)
 
-```bash
-# ALWAYS use patch:
-npm version patch --no-git-tag-version
-```
-
 > **⚠️ ATOMIC COMMIT RULE — Version bump MUST happen before committing feature files.**
 >
 > **CORRECT order:**
@@ -53,7 +68,7 @@ npm version patch --no-git-tag-version
 > This ensures that `git show v2.x.y` always contains both code changes and the version bump together.
 > The GitHub release tag will point to a commit that includes ALL changes for that version.
 
-### 2. Regenerate lock file (REQUIRED after version bump)
+### 3. Regenerate lock file (REQUIRED after version bump)
 
 **Mandatory** — skipping causes `@swc/helpers` lock mismatch and CI failures:
 
@@ -61,7 +76,7 @@ npm version patch --no-git-tag-version
 npm install
 ```
 
-### 3. Finalize CHANGELOG.md
+### 4. Finalize CHANGELOG.md
 
 Replace `[Unreleased]` header with the new version and date.
 Keep an empty `## [Unreleased]` section above it.
@@ -74,7 +89,7 @@ Keep an empty `## [Unreleased]` section above it.
 ## [2.x.y] — YYYY-MM-DD
 ```
 
-### 4. Update openapi.yaml version ⚠️ MANDATORY
+### 5. Update openapi.yaml version ⚠️ MANDATORY
 
 > **CI will fail** if `docs/openapi.yaml` version ≠ `package.json` version (`check:docs-sync` enforces this).
 
@@ -84,33 +99,97 @@ Keep an empty `## [Unreleased]` section above it.
 VERSION=$(node -p "require('./package.json').version") && sed -i "s/  version: .*/  version: $VERSION/" docs/openapi.yaml && echo "✓ openapi.yaml → $VERSION"
 ```
 
-### 5. Stage, commit, and tag
+### 6. Update README.md and i18n docs
+
+Run `/update-docs` workflow steps to:
+
+- Update feature table rows in `README.md`
+- Sync changes to all 29 language `docs/i18n/*/README.md` files
+- Update `docs/FEATURES.md` if Settings section changed
+
+### 7. Run tests
+
+// turbo
+
+```bash
+npm test
+```
+
+All tests must pass before creating the PR.
+
+### 8. Stage, commit, and push
 
 // turbo-all
 
 ```bash
-git add package.json package-lock.json CHANGELOG.md docs/openapi.yaml
+git add -A
 git commit -m "chore(release): v2.x.y — summary of changes"
+git push origin release/v2.x.y
+```
+
+### 9. Open PR to main
+
+```bash
+gh pr create \
+  --repo diegosouzapw/OmniRoute \
+  --base main \
+  --head release/v2.x.y \
+  --title "chore(release): v2.x.y — summary" \
+  --body "## 🚀 Release v2.x.y
+
+### Changes
+...
+
+### Tests
+- X/X tests pass
+
+### ⚠️ After merging: run Phase 2 steps to tag, publish, and deploy."
+```
+
+### 10. 🛑 STOP — Notify User & Await PR Confirmation
+
+**This is a mandatory stop point.** Use `notify_user` with `BlockedOnUser: true`:
+
+Inform the user:
+
+- PR URL
+- Summary of changes
+- Test results
+- List of files changed
+
+**DO NOT proceed to Phase 2 until the user confirms the PR looks good and merges it.**
+
+---
+
+## Phase 2: Post-Merge (only after user confirms)
+
+> Run these steps only AFTER the user has merged the PR.
+
+### 11. Pull main and create tag
+
+```bash
+git checkout main
+git pull origin main
 git tag -a v2.x.y -m "Release v2.x.y"
 ```
 
-### 6. Push to GitHub
+### 12. Push tag to GitHub
 
 ```bash
-git push origin main --tags
+git push origin --tags
 ```
 
-### 7. Create GitHub release
+### 13. Create GitHub release
 
 ```bash
 gh release create v2.x.y --title "v2.x.y — summary" --notes "..."
 ```
 
-### 8. 🐳 Trigger Docker Hub build (MANDATORY — keep npm and Docker in sync)
+### 14. 🐳 Trigger Docker Hub build (MANDATORY — keep npm and Docker in sync)
 
 > **CRITICAL**: Docker Hub and npm MUST always publish the same version.
 > The Docker image is built automatically via GitHub Actions when a new tag is pushed.
-> After pushing the tag in step 5-6, **verify the workflow runs**:
+> After pushing the tag in step 11-12, **verify the workflow runs**:
 
 ```bash
 # Verify the Docker workflow triggered
@@ -129,7 +208,7 @@ If the Docker build was not triggered automatically, trigger it manually:
 gh workflow run docker-publish.yml --repo diegosouzapw/OmniRoute --ref v2.x.y
 ```
 
-### 9. Deploy to BOTH VPS environments (MANDATORY)
+### 15. Deploy to BOTH VPS environments (MANDATORY)
 
 > Always deploy to **both** environments after every release.
 > See `/deploy-vps` workflow for detailed steps.
@@ -151,18 +230,27 @@ curl -s -o /dev/null -w "LOCAL:  HTTP %{http_code}\n" http://192.168.0.15:20128/
 curl -s -o /dev/null -w "AKAMAI: HTTP %{http_code}\n" http://69.164.221.35:20128/
 ```
 
+### 16. Clean up release branch
+
+```bash
+git branch -d release/v2.x.y
+```
+
+---
+
 ## Notes
 
 - Always run `/update-docs` BEFORE this workflow (ensures CHANGELOG and README are current)
 - The `prepublishOnly` script runs `npm run build:cli` automatically during `npm publish`
 - After npm publish, verify with `npm info omniroute version`
 - Lock file sync errors are caused by skipping `npm install` after version bump
+- Use `gh auth switch -u diegosouzapw` if git push fails with wrong account
 
 ## Known CI Pitfalls
 
 | CI failure                                                                | Cause                                                    | Fix                                                                    |
 | ------------------------------------------------------------------------- | -------------------------------------------------------- | ---------------------------------------------------------------------- |
-| `[docs-sync] FAIL - OpenAPI version differs from package.json`            | Skipped step 4 — `docs/openapi.yaml` version not updated | Run step 4 (`sed -i ...`) and commit                                   |
+| `[docs-sync] FAIL - OpenAPI version differs from package.json`            | Skipped step 5 — `docs/openapi.yaml` version not updated | Run step 5 (`sed -i ...`) and commit                                   |
 | `[docs-sync] FAIL - CHANGELOG.md first section must be "## [Unreleased]"` | `## [Unreleased]` missing or not at top of CHANGELOG     | Add `## [Unreleased]\n\n---\n` before the first versioned `## [x.y.z]` |
 | Electron Linux `.deb` build fails (`FpmTarget` error)                     | `fpm` Ruby gem not installed on `ubuntu-latest` runner   | Already fixed in `electron-release.yml` (`gem install fpm` step)       |
 | Docker Hub `502 error writing layer blob`                                 | Transient Docker Hub network error during ARM64 push     | Re-run the Docker publish workflow; no code change needed              |

.gitignore

@@ -89,6 +89,7 @@ docs/*
 !docs/MCP-SERVER.md
 !docs/CLI-TOOLS.md
 
+
 # open-sse tests
 open-sse/test/*
 

CHANGELOG.md

@@ -2,6 +2,345 @@
 
 ## [Unreleased]
 
+> **Coming next** — see [3.0.0-rc branch](https://github.com/diegosouzapw/OmniRoute/tree/3.0.0-rc).
+
+---
+
+## [3.0.0-rc.7] — 2026-03-23 _(What's New vs v2.9.5 — will be released as v3.0.0)_
+
+> **Upgrade from v2.9.5:** 16 issues resolved · 2 community PRs merged · 2 new providers · 7 new API endpoints · 3 new features · DB migration 008+009 · 832 tests passing · 15 sub2api gap improvements (T01–T15 complete).
+
+### 🆕 New Providers
+
+| Provider         | Alias          | Tier | Notes                                                          |
+| ---------------- | -------------- | ---- | -------------------------------------------------------------- |
+| **OpenCode Zen** | `opencode-zen` | Free | 3 models via `opencode.ai/zen/v1` (PR #530 by @kang-heewon)    |
+| **OpenCode Go**  | `opencode-go`  | Paid | 4 models via `opencode.ai/zen/go/v1` (PR #530 by @kang-heewon) |
+
+Both providers use the new `OpencodeExecutor` with multi-format routing (`/chat/completions`, `/messages`, `/responses`, `/models/{model}:generateContent`).
+
+---
+
+### ✨ New Features
+
+#### 🔑 Registered Keys Provisioning API (#464)
+
+Auto-generate and issue OmniRoute API keys programmatically with per-provider and per-account quota enforcement.
+
+| Endpoint                              | Method    | Description                                      |
+| ------------------------------------- | --------- | ------------------------------------------------ |
+| `/api/v1/registered-keys`             | `POST`    | Issue a new key — raw key returned **once only** |
+| `/api/v1/registered-keys`             | `GET`     | List registered keys (masked)                    |
+| `/api/v1/registered-keys/{id}`        | `GET`     | Get key metadata                                 |
+| `/api/v1/registered-keys/{id}`        | `DELETE`  | Revoke a key                                     |
+| `/api/v1/registered-keys/{id}/revoke` | `POST`    | Revoke (for clients without DELETE support)      |
+| `/api/v1/quotas/check`                | `GET`     | Pre-validate quota before issuing                |
+| `/api/v1/providers/{id}/limits`       | `GET/PUT` | Configure per-provider issuance limits           |
+| `/api/v1/accounts/{id}/limits`        | `GET/PUT` | Configure per-account issuance limits            |
+| `/api/v1/issues/report`               | `POST`    | Report quota events to GitHub Issues             |
+
+**DB — Migration 008:** Three new tables: `registered_keys`, `provider_key_limits`, `account_key_limits`.
+**Security:** Keys stored as SHA-256 hashes. Raw key shown once on creation, never retrievable again.
+**Quota types:** `maxActiveKeys`, `dailyIssueLimit`, `hourlyIssueLimit` per provider and per account.
+**Idempotency:** `idempotency_key` field prevents duplicate issuance. Returns `409 IDEMPOTENCY_CONFLICT` if key was already used.
+**Budget per key:** `dailyBudget` / `hourlyBudget` — limits how many requests a key can route per window.
+**GitHub reporting:** Optional. Set `GITHUB_ISSUES_REPO` + `GITHUB_ISSUES_TOKEN` to auto-create GitHub issues on quota exceeded or issuance failures.
+
+#### 🎨 Provider Icons — @lobehub/icons (#529)
+
+All provider icons in the dashboard now use `@lobehub/icons` React components (130+ providers with SVG).
+Fallback chain: **Lobehub SVG → existing `/providers/{id}.png` → generic icon**. Uses a proper React `ErrorBoundary` pattern.
+
+#### 🔄 Model Auto-Sync Scheduler (#488)
+
+OmniRoute now automatically refreshes model lists for connected providers every **24 hours**.
+
+- Runs on server startup via the existing `/api/sync/initialize` hook
+- Configurable via `MODEL_SYNC_INTERVAL_HOURS` environment variable
+- Covers 16 major providers
+- Records last sync time in the settings database
+
+---
+
+### 🔧 Bug Fixes
+
+#### OAuth & Auth
+
+- **#537 — Gemini CLI OAuth:** Clear actionable error when `GEMINI_OAUTH_CLIENT_SECRET` is missing in Docker/self-hosted deployments. Previously showed cryptic `client_secret is missing` from Google. Now provides specific `docker-compose.yml` and `~/.omniroute/.env` instructions.
+
+#### Providers & Routing
+
+- **#536 — LongCat AI:** Fixed `baseUrl` (`api.longcat.chat/openai`) and `authHeader` (`Authorization: Bearer`).
+- **#535 — Pinned model override:** `body.model` is now correctly set to `pinnedModel` when context-cache protection is active.
+- **#532 — OpenCode Go key validation:** Now uses the `zen/v1` test endpoint (`testKeyBaseUrl`) — same key works for both tiers.
+
+#### CLI & Tools
+
+- **#527 — Claude Code + Codex loop:** `tool_result` blocks are now converted to text instead of dropped, stopping infinite tool-result loops.
+- **#524 — OpenCode config save:** Added `saveOpenCodeConfig()` handler (XDG_CONFIG_HOME aware, writes TOML).
+- **#521 — Login stuck:** Login no longer freezes after skipping password setup — redirects correctly to onboarding.
+- **#522 — API Manager:** Removed misleading "Copy masked key" button (replaced with a lock icon tooltip).
+- **#532 — OpenCode Go config:** Guide settings handler now handles `opencode` toolId.
+
+#### Developer Experience
+
+- **#489 — Antigravity:** Missing `googleProjectId` returns a structured 422 error with reconnect guidance instead of a cryptic crash.
+- **#510 — Windows paths:** MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\\Program Files\\...` automatically.
+- **#492 — CLI startup:** `omniroute` CLI now detects `mise`/`nvm`-managed Node when `app/server.js` is missing and shows targeted fix instructions.
+
+---
+
+### 📖 Documentation Updates
+
+- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented
+- **#520** — pnpm: `pnpm approve-builds better-sqlite3` step documented
+
+---
+
+### ✅ Issues Resolved in v3.0.0
+
+`#464` `#488` `#489` `#492` `#510` `#513` `#520` `#521` `#522` `#524` `#527` `#529` `#532` `#535` `#536` `#537`
+
+---
+
+### 🔀 Community PRs Merged
+
+| PR       | Author       | Summary                                                                |
+| -------- | ------------ | ---------------------------------------------------------------------- |
+| **#530** | @kang-heewon | OpenCode Zen + Go providers with `OpencodeExecutor` and improved tests |
+
+---
+
+## [3.0.0-rc.7] - 2026-03-23
+
+### 🔧 Improvements (sub2api Gap Analysis — T05, T08, T09, T13, T14)
+
+- **T05** — Rate-limit DB persistence: `setConnectionRateLimitUntil()`, `isConnectionRateLimited()`, `getRateLimitedConnections()` in `providers.ts`. The existing `rate_limited_until` column is now exposed as a dedicated API — OAuth token refresh must NOT touch this field to prevent rate-limit loops.
+- **T08** — Per-API-key session limit: `max_sessions INTEGER DEFAULT 0` added to `api_keys` via auto-migration. `sessionManager.ts` gains `registerKeySession()`, `unregisterKeySession()`, `checkSessionLimit()`, and `getActiveSessionCountForKey()`. Callers in `chatCore.js` can enforce the limit and decrement on `req.close`.
+- **T09** — Codex vs Spark rate-limit scopes: `getCodexModelScope()` and `getCodexRateLimitKey()` in `codex.ts`. Standard models (`gpt-5.x-codex`, `codex-mini`) get scope `"codex"`; spark models (`codex-spark*`) get scope `"spark"`. Rate-limit keys should be `${accountId}:${scope}` so exhausting one pool doesn't block the other.
+- **T13** — Stale quota display fix: `getEffectiveQuotaUsage(used, resetAt)` returns `0` when the reset window has passed; `formatResetCountdown(resetAt)` returns a human-readable countdown string (e.g. `"2h 35m"`). Both exported from `providers.ts` + `localDb.ts` for dashboard consumption.
+- **T14** — Proxy fast-fail: new `src/lib/proxyHealth.ts` with `isProxyReachable(proxyUrl, timeoutMs=2000)` (TCP check, ≤2s instead of 30s timeout), `getCachedProxyHealth()`, `invalidateProxyHealth()`, and `getAllProxyHealthStatuses()`. Results cached 30s by default; configurable via `PROXY_FAST_FAIL_TIMEOUT_MS` / `PROXY_HEALTH_CACHE_TTL_MS`.
+
+### 🧪 Tests
+
+- Test suite: **832 tests, 0 failures**
+
+---
+
+## [3.0.0-rc.6] - 2026-03-23
+
+### 🔧 Bug Fixes & Improvements (sub2api Gap Analysis — T01–T15)
+
+- **T01** — `requested_model` column in `call_logs` (migration 009): track which model the client originally requested vs the actual routed model. Enables fallback rate analytics.
+- **T02** — Strip empty text blocks from nested `tool_result.content`: prevents Anthropic 400 errors (`text content blocks must be non-empty`) when Claude Code chains tool results.
+- **T03** — Parse `x-codex-5h-*` / `x-codex-7d-*` headers: `parseCodexQuotaHeaders()` + `getCodexResetTime()` extract Codex quota windows for precise cooldown scheduling instead of generic 5-min fallback.
+- **T04** — `X-Session-Id` header for external sticky routing: `extractExternalSessionId()` in `sessionManager.ts` reads `x-session-id` / `x-omniroute-session` headers with `ext:` prefix to avoid collision with internal SHA-256 session IDs. Nginx-compatible (hyphenated header).
+- **T06** — Account deactivated → permanent block: `isAccountDeactivated()` in `accountFallback.ts` detects 401 deactivation signals and applies a 1-year cooldown to prevent retrying permanently dead accounts.
+- **T07** — X-Forwarded-For IP validation: new `src/lib/ipUtils.ts` with `extractClientIp()` and `getClientIpFromRequest()` — skips `unknown`/non-IP entries in `X-Forwarded-For` chains (Nginx/proxy-forwarded requests).
+- **T10** — Credits exhausted → distinct fallback: `isCreditsExhausted()` in `accountFallback.ts` returns 1h cooldown with `creditsExhausted` flag, distinct from generic 429 rate limiting.
+- **T11** — `max` reasoning effort → 131072 budget tokens: `EFFORT_BUDGETS` and `THINKING_LEVEL_MAP` updated; reverse mapping now returns `"max"` for full-budget responses. Unit test updated.
+- **T12** — MiniMax M2.7 pricing entries added: `minimax-m2.7`, `MiniMax-M2.7`, `minimax-m2.7-highspeed` added to pricing table (sub2api PR #1120). M2.5/GLM-4.7/GLM-5/Kimi pricing already existed.
+- **T15** — Array content normalization: `normalizeContentToString()` helper in `openai-to-claude.ts` correctly collapses array-formatted system/tool messages to string before sending to Anthropic.
+
+### 🧪 Tests
+
+- Test suite: **832 tests, 0 failures** (unchanged from rc.5)
+
+---
+
+## [3.0.0-rc.5] - 2026-03-22
+
+### ✨ New Features
+
+- **#464** — Registered Keys Provisioning API: auto-issue API keys with per-provider & per-account quota enforcement
+  - `POST /api/v1/registered-keys` — issue keys with idempotency support
+  - `GET /api/v1/registered-keys` — list (masked) registered keys
+  - `GET /api/v1/registered-keys/{id}` — get key metadata
+  - `DELETE /api/v1/registered-keys/{id}` / `POST ../{id}/revoke` — revoke keys
+  - `GET /api/v1/quotas/check` — pre-validate before issuing
+  - `PUT /api/v1/providers/{id}/limits` — set provider issuance limits
+  - `PUT /api/v1/accounts/{id}/limits` — set account issuance limits
+  - `POST /api/v1/issues/report` — optional GitHub issue reporting
+  - DB migration 008: `registered_keys`, `provider_key_limits`, `account_key_limits` tables
+
+---
+
+## [3.0.0-rc.4] - 2026-03-22
+
+### ✨ New Features
+
+- **#530 (PR)** — OpenCode Zen and OpenCode Go providers added (by @kang-heewon)
+  - New `OpencodeExecutor` with multi-format routing (`/chat/completions`, `/messages`, `/responses`)
+  - 7 models across both tiers
+
+---
+
+## [3.0.0-rc.3] - 2026-03-22
+
+### ✨ New Features
+
+- **#529** — Provider icons now use [@lobehub/icons](https://github.com/lobehub/lobe-icons) with graceful PNG fallback and a `ProviderIcon` component (130+ providers supported)
+- **#488** — Auto-update model lists every 24h via `modelSyncScheduler` (configurable via `MODEL_SYNC_INTERVAL_HOURS`)
+
+### 🔧 Bug Fixes
+
+- **#537** — Gemini CLI OAuth: now shows clear actionable error when `GEMINI_OAUTH_CLIENT_SECRET` is missing in Docker/self-hosted deployments
+
+---
+
+## [3.0.0-rc.2] - 2026-03-22
+
+### 🔧 Bug Fixes
+
+- **#536** — LongCat AI key validation: fixed baseUrl (`api.longcat.chat/openai`) and authHeader (`Authorization: Bearer`)
+- **#535** — Pinned model override: `body.model` is now set to `pinnedModel` when context-cache protection detects a pinned model
+- **#524** — OpenCode config now saved correctly: added `saveOpenCodeConfig()` handler (XDG_CONFIG_HOME aware, writes TOML)
+
+---
+
+## [3.0.0-rc.1] - 2026-03-22
+
+### 🔧 Bug Fixes
+
+- **#521** — Login no longer gets stuck after skipping password setup (redirects to onboarding)
+- **#522** — API Manager: Removed misleading "Copy masked key" button (replaced with lock icon tooltip)
+- **#527** — Claude Code + Codex superpowers loop: `tool_result` blocks now converted to text instead of dropped
+- **#532** — OpenCode GO API key validation now uses the correct `zen/v1` endpoint (`testKeyBaseUrl`)
+- **#489** — Antigravity: missing `googleProjectId` returns structured 422 error with reconnect guidance
+- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\\Program Files\\...`
+- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix
+
+### 📖 Documentation
+
+- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented
+- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented
+
+### ✅ Closed Issues
+
+#489, #492, #510, #513, #520, #521, #522, #525, #527, #532
+
+---
+
+## [2.9.5] — 2026-03-22
+
+> Sprint: New OpenCode providers, embedding credentials fix, CLI masked key bug, CACHE_TAG_PATTERN fix.
+
+### 🐛 Bug Fixes
+
+- **CLI tools save masked API key to config files** — `claude-settings`, `cline-settings`, and `openclaw-settings` POST routes now accept a `keyId` param and resolve the real API key from DB before writing to disk. `ClaudeToolCard` updated to send `keyId` instead of the masked display string. Fixes #523, #526.
+- **Custom embedding providers: `No credentials` error** — `/v1/embeddings` now tracks `credentialsProviderId` separately from the routing prefix, so credentials are fetched from the matching provider node ID rather than the public prefix string. Fixes a regression where `google/gemini-embedding-001` and similar custom-provider models would always fail with a credentials error. Fixes #532-related. (PR #528 by @jacob2826)
+- **Context cache protection regex misses `\n` prefix** — `CACHE_TAG_PATTERN` in `comboAgentMiddleware.ts` updated to match both literal `\n` (backslash-n) and actual newline U+000A that `combo.ts` streaming injects around the `<omniModel>` tag after fix #515. Fixes #531.
+
+### ✨ New Providers
+
+- **OpenCode Zen** — Free tier gateway at `opencode.ai/zen/v1` with 3 models: `minimax-m2.5-free`, `big-pickle`, `gpt-5-nano`
+- **OpenCode Go** — Subscription service at `opencode.ai/zen/go/v1` with 4 models: `glm-5`, `kimi-k2.5`, `minimax-m2.7` (Claude format), `minimax-m2.5` (Claude format)
+- Both providers use the new `OpencodeExecutor` which routes dynamically to `/chat/completions`, `/messages`, `/responses`, or `/models/{model}:generateContent` based on the requested model. (PR #530 by @kang-heewon)
+
+---
+
+## [2.9.4] — 2026-03-21
+
+> Sprint: Bug fixes — preserve Codex prompt cache key, fix tagContent JSON escaping, sync expired token status to DB.
+
+### 🐛 Bug Fixes
+
+- **fix(translator)**: Preserve `prompt_cache_key` in Responses API → Chat Completions translation (#517)
+  — The field is a cache-affinity signal used by Codex; stripping it was preventing prompt cache hits.
+  Fixed in `openai-responses.ts` and `responsesApiHelper.ts`.
+
+- **fix(combo)**: Escape `\n` in `tagContent` so injected JSON string is valid (#515)
+  — Template literal newlines (U+000A) are not allowed unescaped inside JSON string values.
+  Replaced with `\\n` literal sequences in `open-sse/services/combo.ts`.
+
+- **fix(usage)**: Sync expired token status back to DB on live auth failure (#491)
+  — When the Limits & Quotas live check returns 401/403, the connection `testStatus` is now updated
+  to `"expired"` in the database so the Providers page reflects the same degraded state.
+  Fixed in `src/app/api/usage/[connectionId]/route.ts`.
+
+---
+
+## [2.9.3] — 2026-03-21
+
+> Sprint: Add 5 new free AI providers — LongCat, Pollinations, Cloudflare AI, Scaleway, AI/ML API.
+
+### ✨ New Providers
+
+- **feat(providers/longcat)**: Add LongCat AI (`lc/`) — 50M tokens/day free (Flash-Lite) + 500K/day (Chat/Thinking) during public beta. OpenAI-compatible, standard Bearer auth.
+- **feat(providers/pollinations)**: Add Pollinations AI (`pol/`) — no API key required. Proxies GPT-5, Claude, Gemini, DeepSeek V3, Llama 4 (1 req/15s free). Custom executor handles optional auth.
+- **feat(providers/cloudflare-ai)**: Add Cloudflare Workers AI (`cf/`) — 10K Neurons/day free (~150 LLM responses or 500s Whisper audio). 50+ models on global edge. Custom executor builds dynamic URL with `accountId` from credentials.
+- **feat(providers/scaleway)**: Add Scaleway Generative APIs (`scw/`) — 1M free tokens for new accounts. EU/GDPR compliant (Paris). Qwen3 235B, Llama 3.1 70B, Mistral Small 3.2.
+- **feat(providers/aimlapi)**: Add AI/ML API (`aiml/`) — $0.025/day free credit, 200+ models (GPT-4o, Claude, Gemini, Llama) via single aggregator endpoint.
+
+### 🔄 Provider Updates
+
+- **feat(providers/together)**: Add `hasFree: true` + 3 permanently free model IDs: `Llama-3.3-70B-Instruct-Turbo-Free`, `Llama-Vision-Free`, `DeepSeek-R1-Distill-Llama-70B-Free`
+- **feat(providers/gemini)**: Add `hasFree: true` + `freeNote` (1,500 req/day, no credit card needed, aistudio.google.com)
+- **chore(providers/gemini)**: Rename display name to `Gemini (Google AI Studio)` for clarity
+
+### ⚙️ Infrastructure
+
+- **feat(executors/pollinations)**: New `PollinationsExecutor` — omits `Authorization` header when no API key provided
+- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials
+- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings
+
+### 📝 Documentation
+
+- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)
+- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables
+- **docs(readme)**: Updated pricing table with 4 new free tier rows
+- **docs(i18n/pt-BR)**: Updated pricing table + added LongCat/Pollinations/Cloudflare AI/Scaleway sections in Portuguese
+- **docs(new-features/ai)**: 10 task spec files + master implementation plan in `docs/new-features/ai/`
+
+### 🧪 Tests
+
+- Test suite: **821 tests, 0 failures** (unchanged)
+
+---
+
+## [2.9.2] — 2026-03-21
+
+> Sprint: Fix media transcription (Deepgram/HuggingFace Content-Type, language detection) and TTS error display.
+
+### 🐛 Bug Fixes
+
+- **fix(transcription)**: Deepgram and HuggingFace audio transcription now correctly map `video/mp4` → `audio/mp4` and other media MIME types via new `resolveAudioContentType()` helper. Previously, uploading `.mp4` files consistently returned "No speech detected" because Deepgram was receiving `Content-Type: video/mp4`.
+- **fix(transcription)**: Added `detect_language=true` to Deepgram requests — auto-detects audio language (Portuguese, Spanish, etc.) instead of defaulting to English. Fixes non-English transcriptions returning empty or garbage results.
+- **fix(transcription)**: Added `punctuate=true` to Deepgram requests for higher-quality transcription output with correct punctuation.
+- **fix(tts)**: `[object Object]` error display in Text-to-Speech responses fixed in both `audioSpeech.ts` and `audioTranscription.ts`. The `upstreamErrorResponse()` function now correctly extracts nested string messages from providers like ElevenLabs that return `{ error: { message: "...", status_code: 401 } }` instead of a flat error string.
+
+### 🧪 Tests
+
+- Test suite: **821 tests, 0 failures** (unchanged)
+
+### Triaged Issues
+
+- **#508** — Tool call format regression: requested proxy logs and provider chain info (`needs-info`)
+- **#510** — Windows CLI healthcheck path: requested shell/Node version info (`needs-info`)
+- **#485** — Kiro MCP tool calls: closed as external Kiro issue (not OmniRoute)
+- **#442** — Baseten /models endpoint: closed (documented manual workaround)
+- **#464** — Key provisioning API: acknowledged as roadmap item
+
+---
+
+## [2.9.1] — 2026-03-21
+
+> Sprint: Fix SSE omniModel data loss, merge per-protocol model compatibility.
+
+### Bug Fixes
+
+- **#511** — Critical: `<omniModel>` tag was sent after `finish_reason:stop` in SSE streams, causing data loss. Tag is now injected into the first non-empty content chunk, guaranteeing delivery before SDKs close the connection.
+
+### Merged PRs
+
+- **PR #512** (@zhangqiang8vip): Per-protocol model compatibility — `normalizeToolCallId` and `preserveOpenAIDeveloperRole` can now be configured per client protocol (OpenAI, Claude, Responses API). New `compatByProtocol` field in model config with Zod validation.
+
+### Triaged Issues
+
+- **#510** — Windows CLI healthcheck_failed: requested PATH/version info
+- **#509** — Turbopack Electron regression: upstream Next.js bug, documented workarounds
+- **#508** — macOS black screen: suggested `--disable-gpu` workaround
+
 ---
 
 ## [2.9.0] — 2026-03-20

README.md

@@ -11,7 +11,9 @@ _Your universal API proxy — one endpoint, 44+ providers, zero downtime. Now wi
 <div align="center">
 
 [![npm version](https://img.shields.io/npm/v/omniroute?color=cb3837&logo=npm)](https://www.npmjs.com/package/omniroute)
+[![npm downloads](https://img.shields.io/npm/dm/omniroute?color=cb3837&logo=npm&label=npm%20downloads)](https://www.npmjs.com/package/omniroute)
 [![Docker Hub](https://img.shields.io/docker/v/diegosouzapw/omniroute?label=Docker%20Hub&logo=docker&color=2496ED)](https://hub.docker.com/r/diegosouzapw/omniroute)
+[![Docker Pulls](https://img.shields.io/docker/pulls/diegosouzapw/omniroute?logo=docker&color=2496ED&label=docker%20pulls)](https://hub.docker.com/r/diegosouzapw/omniroute)
 [![License](https://img.shields.io/github/license/diegosouzapw/OmniRoute)](https://github.com/diegosouzapw/OmniRoute/blob/main/LICENSE)
 [![Website](https://img.shields.io/badge/Website-omniroute.online-blue?logo=google-chrome&logoColor=white)](https://omniroute.online)
 [![WhatsApp](https://img.shields.io/badge/WhatsApp-Community-25D366?logo=whatsapp&logoColor=white)](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t)
@@ -24,6 +26,25 @@ _Your universal API proxy — one endpoint, 44+ providers, zero downtime. Now wi
 
 ---
 
+## 🆕 What's New in v3.0.0
+
+> **Upgrading from v2.9.5?** — See the [full CHANGELOG](CHANGELOG.md#300--2026-03-22-release-candidate--not-yet-merged-to-main) for all changes.
+
+| Area                         | Change                                                                                                                                                                    |
+| ---------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🔑 **Registered Keys API**   | Auto-provision API keys via `POST /api/v1/registered-keys` with per-provider/account quota enforcement, idempotency, SHA-256 storage, and optional GitHub issue reporting |
+| 🎨 **Provider Icons**        | 130+ provider logos via `@lobehub/icons` (SVG) with PNG → generic fallback chain                                                                                          |
+| 🔄 **Model Auto-Sync**       | 24h scheduler refreshes model lists for 16 providers on startup — configurable via `MODEL_SYNC_INTERVAL_HOURS`                                                            |
+| 🌐 **OpenCode Zen/Go**       | Two new providers from @kang-heewon via PR #530: free tier + subscription tier via `OpencodeExecutor`                                                                     |
+| 🐛 **Gemini CLI OAuth**      | Actionable error when `GEMINI_OAUTH_CLIENT_SECRET` is missing in Docker (was cryptic Google error)                                                                        |
+| 🐛 **OpenCode config**       | `saveOpenCodeConfig()` now correctly writes TOML to `XDG_CONFIG_HOME`                                                                                                     |
+| 🐛 **Pinned model override** | `body.model` correctly set to `pinnedModel` on context-cache protection                                                                                                   |
+| 🐛 **Codex/Claude loop**     | `tool_result` blocks now converted to text to stop infinite loops                                                                                                         |
+| 🐛 **Login redirect**        | Login no longer freezes after skipping password setup                                                                                                                     |
+| 🐛 **Windows paths**         | MSYS2/Git-Bash paths (`/c/...`) normalized to `C:\...` automatically                                                                                                      |
+
+---
+
 ## 🖼️ Main Dashboard
 
 <div align="center">
@@ -716,7 +737,7 @@ Outcome: deep fallback depth for deadline-critical workloads
 
 **Point any IDE/CLI to:** `http://localhost:20128/v1` · API Key: `any-string` · Done.
 
-> **Optional extra coverage (also free):** Groq API key (30 RPM free), NVIDIA NIM (40 RPM free, 70+ models), Cerebras (1M tok/day).
+> **Optional extra coverage (also free):** Groq API key (30 RPM free), NVIDIA NIM (40 RPM free, 70+ models), Cerebras (1M tok/day), LongCat API key (50M tokens/day!), Cloudflare Workers AI (10K Neurons/day, 50+ models).
 
 ## ⚡ Quick Start
 
@@ -921,18 +942,28 @@ When minimized, OmniRoute lives in your system tray with quick actions:
 | **🆓 FREE**         | iFlow                       | **$0**                    | Unlimited        | 5 models unlimited                |
 |                     | Qwen                        | **$0**                    | Unlimited        | 4 models unlimited                |
 |                     | Kiro                        | **$0**                    | Unlimited        | Claude Sonnet/Haiku (AWS Builder) |
+|                     | LongCat Flash-Lite 🆕       | **$0** (50M tok/day 🔥)   | 1 RPS            | Largest free quota on Earth       |
+|                     | Pollinations AI 🆕          | **$0** (no key needed)    | 1 req/15s        | GPT-5, Claude, DeepSeek, Llama 4  |
+|                     | Cloudflare Workers AI 🆕    | **$0** (10K Neurons/day)  | ~150 resp/day    | 50+ models, global edge           |
+|                     | Scaleway AI 🆕              | **$0** (1M tokens total)  | Rate limited     | EU/GDPR, Qwen3 235B, Llama 70B    |
 
 > 🆕 **New models added (Mar 2026):** Grok-4 Fast family at $0.20/$0.50/M (benchmarked at 1143ms — 30% faster than Gemini 2.5 Flash), GLM-5 via Z.AI with 128K output, MiniMax M2.5 reasoning, DeepSeek V3.2 updated pricing, Kimi K2.5 via Moonshot direct API.
 
 **💡 $0 Combo Stack — The Complete Free Setup:**
 
 ```
-Gemini CLI (180K/mo free)
-  → iFlow (unlimited: kimi-k2-thinking, qwen3-coder-plus, deepseek-r1)
-  → Kiro (Claude Sonnet 4.5 + Haiku — unlimited, via AWS Builder ID)
-  → Qwen (4 models — unlimited)
-  → Groq (14.4K req/day — ultra-fast)
-  → NVIDIA NIM (70+ models — 40 RPM forever)
+# 🆓 Ultimate Free Stack 2026 — 11 Providers, $0 Forever
+Kiro (kr/)             → Claude Sonnet/Haiku UNLIMITED
+iFlow (if/)            → kimi-k2-thinking, qwen3-coder-plus, deepseek-r1 UNLIMITED
+LongCat Lite (lc/)     → LongCat-Flash-Lite — 50M tokens/day 🔥
+Pollinations (pol/)    → GPT-5, Claude, DeepSeek, Llama 4 — no key needed
+Qwen (qw/)             → qwen3-coder-plus, qwen3-coder-flash, qwen3-coder-next UNLIMITED
+Gemini (gemini/)       → Gemini 2.5 Flash — 1,500 req/day free API key
+Cloudflare AI (cf/)    → Llama 70B, Gemma 3, Mistral — 10K Neurons/day
+Scaleway (scw/)        → Qwen3 235B, Llama 70B — 1M free tokens (EU)
+Groq (groq/)           → Llama/Gemma ultra-fast — 14.4K req/day
+NVIDIA NIM (nvidia/)   → 70+ open models — 40 RPM forever
+Cerebras (cerebras/)   → Llama/Qwen world-fastest — 1M tok/day
 ```
 
 **Zero cost. Never stops coding.** Configure this as one OmniRoute combo and all fallbacks happen automatically — no manual switching ever.
@@ -1003,19 +1034,66 @@ Available free: `llama-3.3-70b`, `llama-3.1-8b`, `deepseek-r1-distill-llama-70b`
 
 Available free: `llama-3.3-70b-versatile`, `gemma2-9b-it`, `mixtral-8x7b`, `whisper-large-v3`
 
-> **💡 The Ultimate Free Stack:**
+### 🔴 LONGCAT AI (Free API Key — longcat.chat) 🆕
+
+| Model                         | Prefix | Daily Free Quota  | Notes                   |
+| ----------------------------- | ------ | ----------------- | ----------------------- |
+| `LongCat-Flash-Lite`          | `lc/`  | **50M tokens** 💥 | Largest free quota ever |
+| `LongCat-Flash-Chat`          | `lc/`  | 500K tokens       | Multi-turn chat         |
+| `LongCat-Flash-Thinking`      | `lc/`  | 500K tokens       | Reasoning / CoT         |
+| `LongCat-Flash-Thinking-2601` | `lc/`  | 500K tokens       | Jan 2026 version        |
+| `LongCat-Flash-Omni-2603`     | `lc/`  | 500K tokens       | Multimodal              |
+
+> 100% free while in public beta. Sign up at [longcat.chat](https://longcat.chat) with email or phone. Resets daily 00:00 UTC.
+
+### 🟢 POLLINATIONS AI (No API Key Required) 🆕
+
+| Model      | Prefix | Rate Limit | Provider Behind    |
+| ---------- | ------ | ---------- | ------------------ |
+| `openai`   | `pol/` | 1 req/15s  | GPT-5              |
+| `claude`   | `pol/` | 1 req/15s  | Anthropic Claude   |
+| `gemini`   | `pol/` | 1 req/15s  | Google Gemini      |
+| `deepseek` | `pol/` | 1 req/15s  | DeepSeek V3        |
+| `llama`    | `pol/` | 1 req/15s  | Meta Llama 4 Scout |
+| `mistral`  | `pol/` | 1 req/15s  | Mistral AI         |
+
+> ✨ **Zero friction:** No signup, no API key. Add the Pollinations provider with an empty key field and it works immediately.
+
+### 🟠 CLOUDFLARE WORKERS AI (Free API Key — cloudflare.com) 🆕
+
+| Tier | Daily Neurons | Equivalent Usage                        | Notes                   |
+| ---- | ------------- | --------------------------------------- | ----------------------- |
+| Free | **10,000**    | ~150 LLM resp / 500s audio / 15K embeds | Global edge, 50+ models |
+
+Popular free models: `@cf/meta/llama-3.3-70b-instruct`, `@cf/google/gemma-3-12b-it`, `@cf/openai/whisper-large-v3-turbo` (free audio!), `@cf/qwen/qwen2.5-coder-15b-instruct`
+
+> Requires API Token + Account ID from [dash.cloudflare.com](https://dash.cloudflare.com). Store Account ID in provider settings.
+
+### 🟣 SCALEWAY AI (1M Free Tokens — scaleway.com) 🆕
+
+| Tier | Free Quota    | Location     | Notes                               |
+| ---- | ------------- | ------------ | ----------------------------------- |
+| Free | **1M tokens** | 🇫🇷 Paris, EU | No credit card needed within limits |
+
+Available free: `qwen3-235b-a22b-instruct-2507` (Qwen3 235B!), `llama-3.1-70b-instruct`, `mistral-small-3.2-24b-instruct-2506`, `deepseek-v3-0324`
+
+> EU/GDPR compliant. Get API key at [console.scaleway.com](https://console.scaleway.com).
+
+> **💡 The Ultimate Free Stack (11 Providers, $0 Forever):**
 >
 > ```
-> Kiro (Claude, unlimited)
->  → iFlow (5 models, unlimited)
->  → Qwen (4 models, unlimited)
->  → Gemini CLI (180K/mo)
->  → Cerebras (1M tok/day)
->  → Groq (14.4K req/day)
->  → NVIDIA NIM (40 RPM, 70+ models)
+> Kiro (kr/)             → Claude Sonnet/Haiku UNLIMITED
+> iFlow (if/)            → kimi-k2-thinking, qwen3-coder-plus, deepseek-r1 UNLIMITED
+> LongCat Lite (lc/)     → LongCat-Flash-Lite — 50M tokens/day 🔥
+> Pollinations (pol/)    → GPT-5, Claude, DeepSeek, Llama 4 — no key needed
+> Qwen (qw/)             → qwen3-coder models UNLIMITED
+> Gemini (gemini/)       → Gemini 2.5 Flash — 1,500 req/day free
+> Cloudflare AI (cf/)    → 50+ models — 10K Neurons/day
+> Scaleway (scw/)        → Qwen3 235B, Llama 70B — 1M free tokens (EU)
+> Groq (groq/)           → Llama/Gemma — 14.4K req/day ultra-fast
+> NVIDIA NIM (nvidia/)   → 70+ open models — 40 RPM forever
+> Cerebras (cerebras/)   → Llama/Qwen world-fastest — 1M tok/day
 > ```
->
-> Configure this as an OmniRoute combo and you'll never pay for AI again.
 
 ## 🎙️ Free Transcription Combo
 
@@ -1105,17 +1183,17 @@ OmniRoute v2.0 is built as an operational platform, not just a relay proxy.
 
 ### 🎵 Multi-Modal APIs
 
-| Feature                    | What It Does                                                                                                 |
-| -------------------------- | ------------------------------------------------------------------------------------------------------------ |
-| 🖼️ **Image Generation**    | `/v1/images/generations` with cloud and local backends                                                       |
-| 📐 **Embeddings**          | `/v1/embeddings` for search and RAG pipelines                                                                |
-| 🎤 **Audio Transcription** | `/v1/audio/transcriptions` (Whisper and additional providers)                                                |
-| 🔊 **Text-to-Speech**      | `/v1/audio/speech` (multiple engines/providers)                                                              |
-| 🎬 **Video Generation**    | `/v1/videos/generations` (ComfyUI + SD WebUI workflows)                                                      |
-| 🎵 **Music Generation**    | `/v1/music/generations` (ComfyUI workflows)                                                                  |
-| 🛡️ **Moderations**         | `/v1/moderations` safety checks                                                                              |
-| 🔀 **Reranking**           | `/v1/rerank` for relevance scoring                                                                           |
-| 🔍 **Web Search** 🆕       | `/v1/search` — 5 providers (Serper, Brave, Perplexity, Exa, Tavily), 6,500+ free/month, auto-failover, cache |
+| Feature                    | What It Does                                                                                                                                                               |
+| -------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Image Generation**    | `/v1/images/generations` with cloud and local backends                                                                                                                     |
+| 📐 **Embeddings**          | `/v1/embeddings` for search and RAG pipelines                                                                                                                              |
+| 🎤 **Audio Transcription** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Text-to-Speech**      | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise) with correct error messages          |
+| 🎬 **Video Generation**    | `/v1/videos/generations` (ComfyUI + SD WebUI workflows)                                                                                                                    |
+| 🎵 **Music Generation**    | `/v1/music/generations` (ComfyUI workflows)                                                                                                                                |
+| 🛡️ **Moderations**         | `/v1/moderations` safety checks                                                                                                                                            |
+| 🔀 **Reranking**           | `/v1/rerank` for relevance scoring                                                                                                                                         |
+| 🔍 **Web Search** 🆕       | `/v1/search` — 5 providers (Serper, Brave, Perplexity, Exa, Tavily), 6,500+ free/month, auto-failover, cache                                                               |
 
 ### 🛡️ Resilience, Security & Governance
 

ZWS_README_V5.md

@@ -0,0 +1,332 @@
+# ZWS_README_V5 — 按协议配置模型兼容性 + 前端性能优化
+
+V4 内容（HMR 泄漏修复、Edge 警告消除、测试稳定性）已完成；V5 在 V4 基础上实现**按协议维度配置模型兼容性**，新增前端查找性能优化与类型安全改进。
+
+---
+
+## 一、如何发现问题
+
+### 现象
+
+- 同一模型被 **OpenAI Chat Completions**、**OpenAI Responses API**、**Anthropic Messages** 三种客户端请求形态调用时，V2 的兼容性开关（工具 ID 9 位、不保留 developer 角色）是**全局生效**的——无法为不同协议设置不同的兼容策略。
+- 例如：用户希望 OpenAI Responses API 请求时不保留 developer 角色（MiniMax 422 修复），但 OpenAI Chat Completions 请求时保留。V2 下只能二选一。
+- 前端兼容性弹层未标明当前配置对应哪种协议，容易误导。
+- 前端组件中 `Array.find()` 在每次渲染时对 customModels 和 modelCompatOverrides 做 O(n) 线性扫描，模型数量多时存在不必要的性能开销。
+- `ModelCompatPatch` 类型定义与运行时逻辑不一致：`preserveOpenAIDeveloperRole` 字段需要支持 `null`（表示取消设置/恢复默认），但类型仅允许 `boolean`。
+
+### 排查过程
+
+1. **需求分析**：梳理 `detectFormat(body)` 返回的三种协议键（`openai`、`openai-responses`、`claude`），确认每种协议对 developer 角色和 tool call ID 的需求不同。
+2. **数据模型设计**：在现有 `normalizeToolCallId` / `preserveOpenAIDeveloperRole` 顶层字段基础上，设计 `compatByProtocol` 嵌套结构，按协议键细分。
+3. **构建问题**：客户端 `"use client"` 组件直接从 `@/lib/localDb` 引入常量时，间接拉入了 `node:crypto`（经由 `db/proxies.ts`），触发 Webpack `UnhandledSchemeError`。需将常量拆到 `shared/` 层。
+4. **前端性能**：通过 React DevTools 和代码审计发现 `effectiveNormalizeForProtocol` 等函数每次调用都对数组做 `find()`，在渲染列表时存在 O(n²) 的隐患。
+
+---
+
+## 二、根因分析
+
+### 根因 1（P0）：兼容选项无协议维度
+
+V2 的 `normalizeToolCallId` / `preserveOpenAIDeveloperRole` 存储在模型级别的顶层字段，无法区分请求来源协议。`chatCore.ts` 中的 getter 函数只接收 `(providerId, modelId)` 两个参数，不感知当前请求的 `sourceFormat`。
+
+**影响**：跨协议场景下用户只能设置一个全局值，无法精确控制。
+
+### 根因 2（P1）：客户端构建拉入 Node.js 模块
+
+`page.tsx`（"use client"）→ `@/lib/localDb` → `db/proxies.ts` → `import { randomUUID } from "node:crypto"`
+
+Webpack 无法处理 `node:` URI scheme，报 `UnhandledSchemeError`。虽然 V4 已将 `node:crypto` → `crypto` 修复了 `proxies.ts`，但 `localDb.ts` 的 barrel export 链仍然存在风险——客户端组件不应引入任何可能传递到 Node.js 模块的路径。
+
+### 根因 3（P2）：前端查找性能
+
+`effectiveNormalizeForProtocol`、`effectivePreserveForProtocol`、`anyNormalizeCompatBadge`、`anyNoPreserveCompatBadge` 四个函数每次调用都使用 `Array.find()` 在 `customModels` 和 `modelCompatOverrides` 数组中查找目标模型。在模型列表渲染时，每个模型行会调用多次这些函数，导致 O(n × m) 的查找开销（n = 模型数，m = 每行调用次数）。
+
+### 根因 4（P2）：类型定义与运行时不一致
+
+```typescript
+// V3 暂存区版本（有问题）
+export type ModelCompatPatch = Partial<
+  Pick<
+    ModelCompatOverride,
+    "normalizeToolCallId" | "preserveOpenAIDeveloperRole" | "compatByProtocol"
+  >
+>;
+```
+
+`ModelCompatOverride.preserveOpenAIDeveloperRole` 类型为 `boolean | undefined`，但 `mergeModelCompatOverride()` 内部有 `=== null` 判断（用于取消设置/恢复默认），类型层面无法覆盖。
+
+---
+
+## 三、修复方案
+
+### 修复 1：`compatByProtocol` 存储与读取（models.ts）
+
+**新增数据结构**：
+
+```typescript
+type CompatByProtocolMap = Partial<Record<ModelCompatProtocolKey, ModelCompatPerProtocol>>;
+
+export type ModelCompatOverride = {
+  id: string;
+  normalizeToolCallId?: boolean;
+  preserveOpenAIDeveloperRole?: boolean;
+  compatByProtocol?: CompatByProtocolMap; // 新增
+};
+```
+
+**读取优先级链**（适用于 `getModelNormalizeToolCallId` 和 `getModelPreserveOpenAIDeveloperRole`）：
+
+```
+compatByProtocol[sourceFormat].field  →  顶层 field  →  默认值
+```
+
+1. 若 `sourceFormat` 属于已知协议键（`openai` / `openai-responses` / `claude`），且 `compatByProtocol[sourceFormat]` 中存在目标字段，使用该值。
+2. 否则回退到顶层字段。
+3. 顶层字段也不存在时使用默认值（normalizeToolCallId=false，preserveOpenAIDeveloperRole=undefined）。
+
+**深度合并逻辑** `deepMergeCompatByProtocol()`：
+
+- 对每个协议键，逐字段合并而非覆盖。
+- `normalizeToolCallId=false` 时删除该字段（不存储 false，减少冗余）。
+- 合并后若整个协议条目为空对象，删除该协议条目。
+- 协议键通过 `isCompatProtocolKey()` 白名单校验，拒绝未知键。
+
+**Getter 签名扩展**（向后兼容，第三参数可选）：
+
+```typescript
+export function getModelNormalizeToolCallId(
+  providerId: string,
+  modelId: string,
+  sourceFormat?: string | null
+): boolean;
+
+export function getModelPreserveOpenAIDeveloperRole(
+  providerId: string,
+  modelId: string,
+  sourceFormat?: string | null
+): boolean | undefined;
+```
+
+**优点**：
+
+- 完全向后兼容：无 `sourceFormat` 参数时行为与 V2 完全一致。
+- 协议键白名单校验防止存储污染。
+- 深度合并保留未变更协议的配置。
+
+**缺点/注意**：
+
+- JSON 存储体积略增（每个模型最多增加 3 个协议条目）。
+- 新增 ~80 行 TypeScript 代码。
+
+### 修复 2：请求管线传入 sourceFormat（chatCore.ts）
+
+```typescript
+const normalizeToolCallId = getModelNormalizeToolCallId(
+  provider || "",
+  model || "",
+  sourceFormat // 新增第三参
+);
+const preserveDeveloperRole = getModelPreserveOpenAIDeveloperRole(
+  provider || "",
+  model || "",
+  sourceFormat // 新增第三参
+);
+```
+
+`sourceFormat` 由已有的 `detectFormat(body)` 返回，无需新增检测逻辑。
+
+**优点**：
+
+- 改动仅 2 行，精准传参。
+- 不影响其他 handler（embeddings、imageGeneration 等不涉及 developer 角色和 tool call ID）。
+
+### 修复 3：API 路由支持 compatByProtocol（route.ts）
+
+**PUT 请求体扩展**：
+
+- 解构 `compatByProtocol` 并传入 `updateCustomModel()`。
+- `compatOnly` 判断扩展：仅含 `provider` + `modelId` + 兼容字段时，走 `mergeModelCompatOverride()` 路径。
+- 使用 `ModelCompatPatch` 类型替代行内类型定义，统一类型来源。
+
+**Zod 校验 schema**：
+
+```typescript
+const modelCompatPerProtocolSchema = z.object({
+  normalizeToolCallId: z.boolean().optional(),
+  preserveOpenAIDeveloperRole: z.boolean().optional(),
+}).strict();  // strict: 拒绝额外字段
+
+compatByProtocol: z
+  .record(z.enum(["openai", "openai-responses", "claude"]), modelCompatPerProtocolSchema)
+  .optional(),
+```
+
+**优点**：
+
+- `.strict()` 防止客户端注入额外字段。
+- `z.enum()` 限定协议键，与后端白名单一致。
+- 仅传 `compatByProtocol` 即可更新，前端无需拼装完整模型对象。
+
+### 修复 4：客户端安全常量拆分（modelCompat.ts）
+
+**新增** `src/shared/constants/modelCompat.ts`：
+
+```typescript
+export const MODEL_COMPAT_PROTOCOL_KEYS = ["openai", "openai-responses", "claude"] as const;
+export type ModelCompatProtocolKey = (typeof MODEL_COMPAT_PROTOCOL_KEYS)[number];
+```
+
+- 不依赖 Node.js / DB 代码，客户端组件可安全引入。
+- `models.ts` 从此模块引入并再导出。
+- `localDb.ts` 新增 `ModelCompatPatch` 类型导出（供 route.ts 使用），不导出协议常量。
+- `page.tsx` 改为从 `@/shared/constants/modelCompat` 引入。
+
+**优点**：
+
+- 彻底切断客户端 → localDb → db → proxies → node:crypto 的依赖链。
+- 协议键定义单一来源（Single Source of Truth）。
+
+### 修复 5：前端协议选择器与按协议解析（page.tsx）
+
+**ModelCompatPopover 重构**：
+
+- 新增协议下拉选择器（`<select>`），可选 OpenAI Chat / OpenAI Responses / Anthropic Messages。
+- 两个开关（工具 ID 9 位、不保留 developer）**针对选中协议**生效。
+- 选择 Claude 协议时隐藏 developer 角色开关（developer 仅对 OpenAI 系有意义）。
+- 保存时以 `{ compatByProtocol: { [protocol]: payload } }` 形式提交，后端按协议合并。
+- 深色模式适配：下拉框使用 `bg-white dark:bg-zinc-800`、`text-zinc-900 dark:text-zinc-100`。
+
+**Props 接口重构**：
+
+旧接口（4 个独立值/回调）：
+
+```typescript
+(normalizeToolCallId, preserveDeveloperRole, onNormalizeChange, onPreserveChange);
+```
+
+新接口（3 个函数式 props）：
+
+```typescript
+effectiveModelNormalize: (protocol: string) => boolean
+effectiveModelPreserveDeveloper: (protocol: string) => boolean
+onCompatPatch: (protocol: string, payload: {...}) => void
+```
+
+所有消费方（`ModelRow`、`PassthroughModelRow`、`CustomModelsSection`、`CompatibleModelsSection`）已同步更新。
+
+**角标显示逻辑**：
+
+- `anyNormalizeCompatBadge()`：任意协议或顶层存在 `normalizeToolCallId=true` 即显示「ID×9」角标。
+- `anyNoPreserveCompatBadge()`：任意协议或顶层存在 `preserveOpenAIDeveloperRole=false` 即显示「不保留」角标。
+
+**CustomModelsSection 增强**：
+
+- 新增 `modelCompatOverrides` 状态，从 API 响应中获取。
+- 新增 `saveCustomCompat()` 函数，支持仅传 `compatByProtocol` 的独立保存。
+
+### 修复 6：前端 Map 查找性能优化（page.tsx）
+
+**问题**：`effectiveNormalizeForProtocol` 等函数对 `customModels` 和 `modelCompatOverrides` 用 `Array.find()` 做 O(n) 查找，在列表渲染时每个模型行多次调用。
+
+**方案**：使用 `useMemo` + `Map` 将数组预建为 O(1) 查找表。
+
+```typescript
+type CompatModelMap = Map<string, CompatModelRow>;
+
+function buildCompatMap(rows: CompatModelRow[]): CompatModelMap {
+  const m = new Map<string, CompatModelRow>();
+  for (const r of rows) if (r.id) m.set(r.id, r);
+  return m;
+}
+
+// 在组件内
+const customMap = useMemo(() => buildCompatMap(modelMeta.customModels), [modelMeta.customModels]);
+const overrideMap = useMemo(
+  () => buildCompatMap(modelMeta.modelCompatOverrides),
+  [modelMeta.modelCompatOverrides]
+);
+```
+
+所有查找函数签名从 `(modelId, protocol, customModels[], overrides[])` 改为 `(modelId, protocol, customMap, overrideMap)`，内部使用 `Map.get()` 替代 `Array.find()`。
+
+**优点**：
+
+- 查找从 O(n) 降为 O(1)。
+- `useMemo` 依赖项正确，仅在数据变化时重建 Map。
+- `CustomModelsSection` 内部也独立构建 Map，不依赖父组件。
+
+### 修复 7：ModelCompatPatch 类型修正（models.ts）
+
+```typescript
+// 修复后 — 显式允许 null
+export type ModelCompatPatch = {
+  normalizeToolCallId?: boolean;
+  preserveOpenAIDeveloperRole?: boolean | null; // null = 取消设置/恢复默认
+  compatByProtocol?: CompatByProtocolMap;
+};
+```
+
+与 `mergeModelCompatOverride()` 内的 `=== null` 判断逻辑一致，类型安全。
+
+### 修复 8：CompatByProtocolMap 类型收紧（page.tsx）
+
+客户端 `CompatByProtocolMap` 从 `Record<string, ...>` 改为 `Record<ModelCompatProtocolKey, ...>`，增强类型安全，防止传入未知协议键。
+
+### 修复 9：i18n 文案新增
+
+| 键名                            | 中文                                          | 英文                                                           |
+| ------------------------------- | --------------------------------------------- | -------------------------------------------------------------- |
+| `compatProtocolLabel`           | 客户端请求协议                                | Client request protocol                                        |
+| `compatProtocolHint`            | 以下选项在 OmniRoute 识别到该请求形态时生效。 | These options apply when OmniRoute detects this request shape. |
+| `compatProtocolOpenAI`          | OpenAI Chat Completions                       | OpenAI Chat Completions                                        |
+| `compatProtocolOpenAIResponses` | OpenAI Responses API                          | OpenAI Responses API                                           |
+| `compatProtocolClaude`          | Anthropic Messages                            | Anthropic Messages                                             |
+
+---
+
+## 四、使用方式
+
+1. 点击模型行的 **「兼容性」** 按钮。
+2. 在弹层内先选择 **「客户端请求协议」**（OpenAI Chat / OpenAI Responses / Anthropic Messages）。
+3. 勾选该协议下的「工具 ID 9 位」或「不保留 developer 角色」。
+4. 保存后，仅在该协议形态的请求下生效。
+5. 未配置某协议时，该协议下行为回退到顶层兼容字段（若存在），再回退到默认值（保留 developer、不规范化 tool id）。
+6. 角标「ID×9」「不保留」在任意协议存在对应配置时显示。
+
+---
+
+## 五、预期效果
+
+| 指标                      | 修复前                | 修复后                                     |
+| ------------------------- | --------------------- | ------------------------------------------ |
+| 兼容性配置维度            | 全局（模型级）        | 按协议（OpenAI Chat / Responses / Claude） |
+| developer 角色精确控制    | 不支持                | 支持（如：仅 Responses API 不保留）        |
+| 前端兼容性查找性能        | O(n) Array.find       | O(1) Map.get（useMemo 缓存）               |
+| ModelCompatPatch 类型安全 | null 值无类型覆盖     | 显式 `boolean \| null`                     |
+| 客户端构建风险            | 可能引入 Node.js 模块 | 已隔离（shared/constants 层）              |
+| API 验证                  | 无 compatByProtocol   | Zod strict schema 校验                     |
+| 深色模式                  | 协议选择器不可读      | bg/text 适配 dark 主题                     |
+
+---
+
+## 六、涉及文件清单
+
+| 区域       | 文件                                                        | 改动类型                                                                                         |
+| ---------- | ----------------------------------------------------------- | ------------------------------------------------------------------------------------------------ |
+| 协议常量   | `src/shared/constants/modelCompat.ts`                       | **新建**，客户端安全的协议键与类型                                                               |
+| 存储与读写 | `src/lib/db/models.ts`                                      | `compatByProtocol` 数据结构、深度合并、getter 第三参 `sourceFormat`、`ModelCompatPatch` 类型修正 |
+| 再导出层   | `src/lib/localDb.ts`                                        | 新增 `ModelCompatPatch` 类型导出                                                                 |
+| API 路由   | `src/app/api/provider-models/route.ts`                      | PUT 支持 `compatByProtocol`，使用 `ModelCompatPatch` 类型                                        |
+| 输入校验   | `src/shared/validation/schemas.ts`                          | `modelCompatPerProtocolSchema`（strict）+ `compatByProtocol` 记录校验                            |
+| 请求管线   | `open-sse/handlers/chatCore.ts`                             | `getModelNormalizeToolCallId` / `getModelPreserveOpenAIDeveloperRole` 传入 `sourceFormat`        |
+| 前端 UI    | `src/app/(dashboard)/dashboard/providers/[id]/page.tsx`     | 协议选择器、按协议解析/保存、角标逻辑、Map 性能优化、类型收紧                                    |
+| i18n       | `src/i18n/messages/en.json`，`src/i18n/messages/zh-CN.json` | 5 条新文案                                                                                       |
+
+---
+
+## 七、回退方案
+
+- **禁用按协议配置**：删除 `compatByProtocol` 字段后，getter 自动回退到顶层字段，行为与 V2 一致。
+- **前端 Map 优化回退**：将 `Map.get()` 改回 `Array.find()` 即可，纯性能优化无功能耦合。
+- **客户端常量回退**：将 `MODEL_COMPAT_PROTOCOL_KEYS` 定义移回 `models.ts` 并从 `localDb.ts` 导出（需同时确保 `node:crypto` 问题不再存在）。
+- **生产环境**：以上修复对生产构建无负面影响。`compatByProtocol` 为可选字段，未配置时默认行为不变。API Zod 校验确保不会接受畸形数据。

bin/omniroute.mjs

@@ -189,8 +189,27 @@ const serverJs = join(APP_DIR, "server.js");
 
 if (!existsSync(serverJs)) {
   console.error("\x1b[31m✖ Server not found at:\x1b[0m", serverJs);
-  console.error("  This usually means the package was not built correctly.");
-  console.error("  Try reinstalling: npm install -g omniroute");
+  console.error("  The package may not have been built correctly.");
+  console.error("");
+  // (#492) Detect common non-standard Node managers that cause this issue
+  const nodeExec = process.execPath || "";
+  const isMise = nodeExec.includes("mise") || nodeExec.includes(".local/share/mise");
+  const isNvm = nodeExec.includes(".nvm") || nodeExec.includes("nvm");
+  if (isMise) {
+    console.error(
+      "  \x1b[33m⚠ mise detected:\x1b[0m If you installed via `npm install -g omniroute`,"
+    );
+    console.error("    try: \x1b[36mnpx omniroute@latest\x1b[0m  (downloads a fresh copy)");
+    console.error("    or:  \x1b[36mmise exec -- npx omniroute\x1b[0m");
+  } else if (isNvm) {
+    console.error(
+      "  \x1b[33m⚠ nvm detected:\x1b[0m Try reinstalling after loading the correct Node version:"
+    );
+    console.error("    \x1b[36mnvm use --lts && npm install -g omniroute\x1b[0m");
+  } else {
+    console.error("  Try: \x1b[36mnpm install -g omniroute\x1b[0m  (reinstall)");
+    console.error("  Or:  \x1b[36mnpx omniroute@latest\x1b[0m");
+  }
   process.exit(1);
 }
 

docs/i18n/ar/README.md

@@ -932,8 +932,8 @@ npm run electron:build:linux   # Linux (.AppImage)
 | ميزة | ماذا يفعل || -------------------------- | ------------------------------------------------------------- |
 | 🖼️ **إنشاء الصور** | `/v1/images/generations` مع الواجهات الخلفية السحابية والمحلية |
 | 📐 **المضامين** | `/v1/embeddings` للبحث وخطوط أنابيب RAG |
-| 🎤 **نسخ صوتي** | `/v1/audio/transcriptions` (مقدمو خدمات الهمس والإضافيون) |
-| 🔊 **تحويل النص إلى كلام** | `/v1/audio/speech` (محركات/موفرو متعددون) |
+| 🎤 **نسخ صوتي** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **تحويل النص إلى كلام** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise) |
 | 🎬 **توليد الفيديو** | `/v1/videos/generations` (سير عمل ComfyUI + SD WebUI) |
 | 🎵 **جيل الموسيقى** | `/v1/music/generations` (سير عمل ComfyUI) |
 | 🛡️ **اعتدالات** | فحوصات السلامة `/v1/moderations` |

docs/i18n/bg/README.md

@@ -933,8 +933,8 @@ OmniRoute v2.0 е създаден като операционна платфо
 | Характеристика | Какво прави || -------------------------- | ------------------------------------------------------------ |
 | 🖼️ **Генериране на изображения** | `/v1/images/generations` с облак и локален бекенд |
 | 📐 **Вграждания** | `/v1/embeddings` за търсене и RAG тръбопроводи |
-| 🎤 **Аудио транскрипция** | `/v1/audio/transcriptions` (Whisper и допълнителни доставчици) |
-| 🔊 **Текст към говор** | `/v1/audio/speech` (множество машини/доставчици) |
+| 🎤 **Аудио транскрипция** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Текст към говор** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise) |
 | 🎬 **Видео генериране** | `/v1/videos/generations` (работни процеси ComfyUI + SD WebUI) |
 | 🎵 **Музикално поколение** | `/v1/music/generations` (работни процеси на ComfyUI) |
 | 🛡️ **Модерации** | `/v1/moderations` проверки за безопасност |

docs/i18n/cs/README.md

@@ -2,7 +2,7 @@
 
 ### Nikdy nepřestávejte s kódováním. Chytré směrování k **BEZPLATNÝM a levným modelům AI** s automatickým přepínáním mezi záložními systémy.
 
-*Váš univerzální API proxy – jeden endpoint, více než 44 poskytovatelů, nulové výpadky. Nyní s orchestrací agentů **MCP a A2A** .*
+_Váš univerzální API proxy – jeden endpoint, více než 44 poskytovatelů, nulové výpadky. Nyní s orchestrací agentů **MCP a A2A** ._
 
 **Dokončení chatu • Vkládání • Generování obrázků • Video • Hudba • Audio • Změna pořadí • **Vyhledávání na webu** • MCP server • A2A protokol • 100% TypeScript**
 
@@ -30,26 +30,23 @@
 <summary><b>Kliknutím zobrazíte snímky obrazovky z řídicího panelu</b></summary>
 </details>
 
-Strana | Snímek obrazovky
---- | ---
-**Poskytovatelé** | ![Poskytovatelé](docs/screenshots/01-providers.png)
-**Kombinace** | ![Kombinace](docs/screenshots/02-combos.png)
-**Analytika** | ![Analytika](docs/screenshots/03-analytics.png)
-**Zdraví** | ![Zdraví](docs/screenshots/04-health.png)
-**Překladatel** | ![Překladatel](docs/screenshots/05-translator.png)
-**Nastavení** | ![Nastavení](docs/screenshots/06-settings.png)
-**Nástroje CLI** | ![Nástroje CLI](docs/screenshots/07-cli-tools.png)
-**Protokoly používání** | ![Používání](docs/screenshots/08-usage.png)
-**Koncové body** | ![Koncové body](docs/screenshots/09-endpoint.png)
-
-
-
+| Strana                  | Snímek obrazovky                                    |
+| ----------------------- | --------------------------------------------------- |
+| **Poskytovatelé**       | ![Poskytovatelé](docs/screenshots/01-providers.png) |
+| **Kombinace**           | ![Kombinace](docs/screenshots/02-combos.png)        |
+| **Analytika**           | ![Analytika](docs/screenshots/03-analytics.png)     |
+| **Zdraví**              | ![Zdraví](docs/screenshots/04-health.png)           |
+| **Překladatel**         | ![Překladatel](docs/screenshots/05-translator.png)  |
+| **Nastavení**           | ![Nastavení](docs/screenshots/06-settings.png)      |
+| **Nástroje CLI**        | ![Nástroje CLI](docs/screenshots/07-cli-tools.png)  |
+| **Protokoly používání** | ![Používání](docs/screenshots/08-usage.png)         |
+| **Koncové body**        | ![Koncové body](docs/screenshots/09-endpoint.png)   |
 
 ---
 
 ### 🤖 Bezplatný poskytovatel umělé inteligence pro vaše oblíbené programátory
 
-*Připojte libovolný nástroj IDE nebo CLI s umělou inteligencí přes OmniRoute — bezplatnou API bránu pro neomezené kódování.*
+_Připojte libovolný nástroj IDE nebo CLI s umělou inteligencí přes OmniRoute — bezplatnou API bránu pro neomezené kódování._
 
   <table>
     <tr>
@@ -68,7 +65,6 @@ Strana | Snímek obrazovky
     </tr>
   </table>
 
-
 <sub>📡 Všichni agenti se připojují přes <code>http://localhost:20128/v1</code> nebo <code>http://cloud.omniroute.online/v1</code> — jedna konfigurace, neomezené modely a kvóty</sub>
 
 ---
@@ -161,9 +157,6 @@ Vývojáři platí za Claude Pro, Codex Pro nebo GitHub Copilot 20–200 dolarů
 - **Vlastní kombinace** — Přizpůsobitelné záložní řetězce se 6 strategiemi vyvažování (fill-first, round robin, P2C, náhodné, nejméně používané, nákladově optimalizované)
 - **Codex Business Quotas** — Sledování kvót pracovního prostoru firmy/týmu přímo v dashboardu
 
-
-
-
 <details>
 <summary><b>🔌 2. „Potřebuji použít více poskytovatelů, ale každý má jiné API“</b></summary>
 </details>
@@ -180,9 +173,6 @@ OpenAI používá jeden formát, Claude (Anthropic) jiný a Gemini ještě třet
 - **Strukturovaný výstup pro Gemini** — `json_schema` → automatická konverze `responseMimeType` / `responseSchema`
 - **Výchozí hodnota `stream` je `false`** – Odpovídá specifikaci OpenAI, čímž se zabrání neočekávanému SSE v Python/Rust/Go SDK.
 
-
-
-
 <details>
 <summary><b>🌐 3. „Můj poskytovatel AI blokuje můj region/zemi“</b></summary>
 </details>
@@ -199,9 +189,6 @@ Poskytovatelé jako OpenAI/Codex blokují přístup z určitých geografických
 - **TLS Fingerprint Spoofing** — Otisk prstu TLS podobný prohlížeči pomocí `wreq-js` pro obcházení detekce botů
 - **🔏 Porovnávání otisků prstů v CLI** — Změní pořadí záhlaví a polí v těle serveru tak, aby odpovídala nativním binárním podpisům v CLI, čímž drasticky snižuje riziko nahlašování účtu. IP adresa proxy je zachována — získáte současně stealth **i** maskování IP adresy.
 
-
-
-
 <details>
 <summary><b>🆓 4. „Chci používat umělou inteligenci pro kódování, ale nemám peníze“</b></summary>
 </details>
@@ -216,9 +203,6 @@ Ne každý si může dovolit zaplatit 20–200 dolarů měsíčně za předplatn
 - **NVIDIA NIM Free Access** — ~40 RPM developerský přístup k více než 70 modelům na build.nvidia.com (přechod z kreditů na čisté limity rychlosti)
 - **Strategie optimalizace nákladů** – Strategie směrování, která automaticky vybere nejlevnějšího dostupného poskytovatele
 
-
-
-
 <details>
 <summary><b>🔒 5. „Potřebuji chránit svou bránu umělé inteligence před neoprávněným přístupem“</b></summary>
 </details>
@@ -236,9 +220,6 @@ Při zpřístupnění brány umělé inteligence síti (LAN, VPS, Docker) může
 - **Ochrana proti vkládání výzev** – Sanitizace proti škodlivým vzorcům výzev
 - **Šifrování AES-256-GCM** – přihlašovací údaje jsou v klidovém stavu šifrovány
 
-
-
-
 <details>
 <summary><b>🛑 6. „Můj poskytovatel selhal a já ztratil/a programovací tok“</b></summary>
 </details>
@@ -254,9 +235,6 @@ Poskytovatelé umělé inteligence se mohou stát nestabilními, vracet chyby 5x
 - **Kombinovaný jistič** – Automaticky deaktivuje selhávajícího poskytovatele v rámci kombinovaného řetězce
 - **Dashboard stavu** — Monitorování provozuschopnosti, stavy jističů, uzamčení, statistiky mezipaměti, latence p50/p95/p99
 
-
-
-
 <details>
 <summary><b>🔧 7. „Konfigurace každého nástroje umělé inteligence je zdlouhavá a opakující se“</b></summary>
 </details>
@@ -270,9 +248,6 @@ Vývojáři používají Cursor, Claude Code, Codex CLI, OpenClaw, Gemini CLI, K
 - **Průvodce zaváděním** – 4krokové nastavení pro začínající uživatele
 - **Jeden koncový bod, všechny modely** – jednou nakonfigurujte `http://localhost:20128/v1` a získejte přístup k více než 44 poskytovatelům
 
-
-
-
 <details>
 <summary><b>🔑 8. „Správa OAuth tokenů od více poskytovatelů je peklo“</b></summary>
 </details>
@@ -288,9 +263,6 @@ Claude Code, Codex, Gemini CLI, Copilot – všechny používají OAuth 2.0 s to
 - **OAuth Behind Nginx** — Používá `window.location.origin` pro kompatibilitu s reverzní proxy
 - **Průvodce vzdáleným OAuth** – Podrobný návod k přihlašovacím údajům Google Cloud na VPS/Dockeru
 
-
-
-
 <details>
 <summary><b>📊 9. „Nevím, kolik utrácím ani kde“</b></summary>
 </details>
@@ -305,9 +277,6 @@ Vývojáři používají více placených poskytovatelů, ale nemají jednotný
 - **Statistiky použití pro každý klíč API** — Počet požadavků a časové razítko posledního použití pro každý klíč
 - **Analytický panel** – Statistické karty, graf využití modelu, tabulka poskytovatelů s mírou úspěšnosti a latencí
 
-
-
-
 <details>
 <summary><b>🐛 10. „Nedokážu diagnostikovat chyby a problémy ve volání umělé inteligence.“</b></summary>
 </details>
@@ -324,9 +293,6 @@ Když volání selže, vývojář neví, zda se jednalo o limit rychlosti, vypr
 - **Souborové protokolování s rotací** – Konzolový interceptor zachycuje vše do protokolu JSON s rotací na základě velikosti
 - **Zpráva o systémových informacích** — příkaz `npm run system-info` vygeneruje `system-info.txt` s kompletním popisem vašeho prostředí (verze uzlu, verze OmniRoute, operační systém, nástroje CLI, stav Dockeru/PM2). Přiložte jej při hlášení problémů pro okamžité třídění.
 
-
-
-
 <details>
 <summary><b>🏗️ 11. „Nasazení a údržba brány je složitá“</b></summary>
 </details>
@@ -343,9 +309,6 @@ Instalace, konfigurace a údržba AI proxy v různých prostředích (lokální,
 - **Cloud Sync** – Konfigurace synchronizace mezi zařízeními pomocí Cloudflare Workers
 - **Zálohy databází** — Automatické zálohování, obnovení, export a import všech nastavení
 
-
-
-
 <details>
 <summary><b>🌍 12. „Rozhraní je pouze v angličtině a můj tým nemluví anglicky“</b></summary>
 </details>
@@ -359,9 +322,6 @@ Týmy v neanglicky mluvících zemích, zejména v Latinské Americe, Asii a Evr
 - **Vícejazyčné soubory README** — 30 kompletních překladů dokumentace
 - **Výběr jazyka** — Ikona glóbu v záhlaví pro přepínání v reálném čase
 
-
-
-
 <details>
 <summary><b>🔄 13. „Potřebuji víc než jen chat – potřebuji vložené soubory, obrázky, zvuk.“</b></summary>
 </details>
@@ -380,9 +340,6 @@ Umělá inteligence není jen dokončování chatu. Vývojáři potřebují gene
 - **Změna pořadí** — `/v1/rerank` — Změna pořadí relevance dokumentu
 - **Responses API** — Plná podpora `/v1/responses` pro Codex
 
-
-
-
 <details>
 <summary><b>🧪 14. „Nemám způsob, jak testovat a porovnávat kvalitu napříč modely.“</b></summary>
 </details>
@@ -397,9 +354,6 @@ Vývojáři chtějí vědět, který model je pro jejich případ použití nejl
 - **Tester chatu** — Kompletní okružní cesta s vizuálním vykreslováním odpovědí
 - **Živý monitor** — Stream všech požadavků procházejících proxy serverem v reálném čase
 
-
-
-
 <details>
 <summary><b>📈 15. „Potřebuji škálovat bez ztráty výkonu“</b></summary>
 </details>
@@ -415,9 +369,6 @@ S rostoucím objemem požadavků generují stejné otázky bez ukládání do me
 - **Mezipaměť pro ověření klíčů API** — třívrstvá mezipaměť pro výkon produkčního prostředí
 - **Dashboard s telemetrií** – latence p50/p95/p99, statistiky mezipaměti, dostupnost
 
-
-
-
 <details>
 <summary><b>🤖 16. „Chci mít chování modelů globálně pod kontrolou“</b></summary>
 </details>
@@ -434,9 +385,6 @@ Vývojáři, kteří chtějí všechny odpovědi v určitém jazyce, se specific
 - **Přepínání poskytovatele** – Povolení/zakázání všech připojení pro poskytovatele jedním kliknutím
 - **Blokovaní poskytovatelé** – Vyloučení konkrétních poskytovatelů ze seznamu `/v1/models`
 
-
-
-
 <details>
 <summary><b>🧰 17. „Potřebuji nástroje MCP jako prvotřídní produktové funkce.“</b></summary>
 </details>
@@ -449,9 +397,6 @@ Mnoho bran umělé inteligence odhaluje MCP pouze jako skrytý implementační d
 - Vyhrazená stránka pro správu MCP s procesy, nástroji, rozsahy a auditem
 - Vestavěný rychlý start pro `omniroute --mcp` a onboarding klienta
 
-
-
-
 <details>
 <summary><b>🧠 18. „Potřebuji orchestraci A2A se synchronizací a cestami úloh streamu.“</b></summary>
 </details>
@@ -464,9 +409,6 @@ Pracovní postupy agentů vyžadují jak přímé odpovědi, tak dlouhodobé str
 - Streamování SSE s šířením stavu terminálu
 - Rozhraní API životního cyklu úloh pro `tasks/get` a `tasks/cancel`
 
-
-
-
 <details>
 <summary><b>🛰️ 19. „Potřebuji skutečný stav procesu MCP, ne odhadovaný stav.“</b></summary>
 </details>
@@ -479,9 +421,6 @@ Provozní týmy potřebují vědět, zda je MCP skutečně aktivní, nejen zda j
 - API stavu MCP kombinující prezenční signál a nedávnou aktivitu
 - Karty stavu uživatelského rozhraní pro zobrazení aktuálnosti procesů/provozuschopnosti/prezenčního signálu
 
-
-
-
 <details>
 <summary><b>📋 20. „Potřebuji auditovatelné provedení nástroje MCP“</b></summary>
 </details>
@@ -494,9 +433,6 @@ Když nástroje mění konfiguraci nebo spouštějí operační akce, týmy pot
 - Filtruje podle nástroje, úspěchu/neúspěchu, klíče API a stránkování
 - Tabulka auditu dashboardu + koncové body statistik pro automatizaci
 
-
-
-
 <details>
 <summary><b>🔐 21. „Potřebuji omezená oprávnění MCP pro každou integraci.“</b></summary>
 </details>
@@ -509,9 +445,6 @@ Různí klienti by měli mít přístup ke kategoriím nástrojů s nejnižším
 - Vynucení rozsahu a viditelnost v uživatelském rozhraní správy MCP
 - Bezpečná výchozí poloha pro provozní nástroje
 
-
-
-
 <details>
 <summary><b>⚙️ 22. „Potřebuji provozní kontroly bez nutnosti přesouvání“</b></summary>
 </details>
@@ -524,9 +457,6 @@ Týmy potřebují rychlé změny v běhovém prostředí během incidentů nebo
 - Používejte profily odolnosti z předdefinovaných balíčků zásad
 - Resetujte stav jističe ze stejného ovládacího panelu
 
-
-
-
 <details>
 <summary><b>🔄 23. „Potřebuji živý přehled o životním cyklu úkolů A2A a jejich zrušení.“</b></summary>
 </details>
@@ -539,9 +469,6 @@ Bez přehledu o životním cyklu je obtížné třídit incidenty úkolů.
 - Podrobný přehled metadat úloh, událostí a artefaktů
 - Koncový bod zrušení úlohy a akce uživatelského rozhraní s potvrzením
 
-
-
-
 <details>
 <summary><b>🌊 24. „Potřebuji metriky aktivního streamu pro A2A zátěž“</b></summary>
 </details>
@@ -554,9 +481,6 @@ Streamovací pracovní postupy vyžadují provozní přehled o souběžnosti a
 - Časové razítko posledního úkolu a počty pro jednotlivé stavy
 - Karty A2A dashboardu pro monitorování provozu v reálném čase
 
-
-
-
 <details>
 <summary><b>🪪 25. „Potřebuji standardní vyhledávání agentů pro klienty“</b></summary>
 </details>
@@ -569,9 +493,6 @@ Externí klienti a orchestratoři potřebují pro onboarding strojově čitelná
 - Schopnosti a dovednosti zobrazené v uživatelském rozhraní pro správu
 - API pro stav A2A zahrnuje metadata pro zjišťování pro automatizaci
 
-
-
-
 <details>
 <summary><b>🧭 26. „Potřebuji v uživatelském rozhraní produktu zjistitelnost protokolu.“</b></summary>
 </details>
@@ -584,9 +505,6 @@ Pokud uživatelé nemohou objevit protokolové povrchy, kvalita přijetí a podp
 - Přepínání stavu inline služby (Online/Offline) pro MCP a A2A
 - Odkazy z přehledu na vyhrazené karty pro správu
 
-
-
-
 <details>
 <summary><b>🧪 27. „Potřebuji komplexní ověření protokolu se skutečnými klienty.“</b></summary>
 </details>
@@ -599,9 +517,6 @@ Simulované testy nestačí k ověření kompatibility protokolu před vydáním
 - Klientské testy A2A pro toky zjišťování, odesílání, streamování, načítání a zrušení
 - Křížová kontrola tvrzení oproti API pro audit MCP a úkoly A2A
 
-
-
-
 <details>
 <summary><b>📡 28. „Potřebuji jednotnou pozorovatelnost napříč všemi rozhraními“</b></summary>
 </details>
@@ -614,9 +529,6 @@ Rozdělení pozorovatelnosti podle protokolu vytváří slepá místa a delší
 - Stav + audit + telemetrie požadavků napříč vrstvami OpenAI, MCP a A2A
 - Provozní API pro stav a automatizaci
 
-
-
-
 <details>
 <summary><b>💼 29. „Potřebuji jeden runtime pro proxy + nástroje + orchestraci agentů“</b></summary>
 </details>
@@ -629,9 +541,6 @@ Spouštění mnoha samostatných služeb zvyšuje provozní náklady a počet po
 - Sdílené ověřování, odolnost, úložiště dat a pozorovatelnost
 - Konzistentní model politik napříč všemi interakčními plochami
 
-
-
-
 <details>
 <summary><b>🚀 30. „Potřebuji agentské pracovní postupy bez slepení kódu.“</b></summary>
 </details>
@@ -644,9 +553,6 @@ Týmy ztrácejí rychlost při spojování více ad-hoc služeb a skriptů.
 - Vestavěná uživatelská rozhraní pro správu protokolů a cesty pro ověřování kouře
 - Základy připravené pro produkční prostředí (zabezpečení, protokolování, odolnost, zálohování)
 
-
-
-
 ### Příklady herních plánů (integrované případy užití)
 
 **Příručka A: Maximalizace placeného předplatného + levné zálohování**
@@ -701,13 +607,13 @@ Outcome: deep fallback depth for deadline-critical workloads
 
 > Nastavte si kódování s umělou inteligencí během několika minut za **0 $/měsíc** . Propojte tyto bezplatné účty a využijte vestavěnou kombinaci **Free Stack** .
 
-Krok | Akce | Poskytovatelé odemčeni
---- | --- | ---
-1 | Připojení **Kiro** (AWS Builder ID OAuth) | Claude Sonnet 4.5, Haiku 4.5 – **neomezeně**
-2 | Připojení k **iFlow** (Google OAuth) | kimi-k2-myšlení, qwen3-coder-plus, deepseek-r1... — **neomezeně**
-3 | Připojení **Qwen** (kód zařízení) | qwen3-coder-plus, qwen3-coder-flash... — **neomezeně**
-4 | Připojení **rozhraní příkazového řádku Gemini** (Google OAuth) | gemini-3-flash, gemini-2.5-pro — **180 000 GBP/měsíc zdarma**
-5 | `/dashboard/combos` → Šablona **Free Stack (0 $)** | Automatické zařazení všech bezplatných poskytovatelů do routingu
+| Krok | Akce                                                           | Poskytovatelé odemčeni                                            |
+| ---- | -------------------------------------------------------------- | ----------------------------------------------------------------- |
+| 1    | Připojení **Kiro** (AWS Builder ID OAuth)                      | Claude Sonnet 4.5, Haiku 4.5 – **neomezeně**                      |
+| 2    | Připojení k **iFlow** (Google OAuth)                           | kimi-k2-myšlení, qwen3-coder-plus, deepseek-r1... — **neomezeně** |
+| 3    | Připojení **Qwen** (kód zařízení)                              | qwen3-coder-plus, qwen3-coder-flash... — **neomezeně**            |
+| 4    | Připojení **rozhraní příkazového řádku Gemini** (Google OAuth) | gemini-3-flash, gemini-2.5-pro — **180 000 GBP/měsíc zdarma**     |
+| 5    | `/dashboard/combos` → Šablona **Free Stack (0 $)**             | Automatické zařazení všech bezplatných poskytovatelů do routingu  |
 
 **V libovolném IDE/CLI naveďte:** `http://localhost:20128/v1` · Klíč API: `any-string` · Hotovo.
 
@@ -732,13 +638,13 @@ omniroute
 
 Dashboard se otevírá na `http://localhost:20128` a základní URL API je `http://localhost:20128/v1` .
 
-Příkaz | Popis
---- | ---
-`omniroute` | Spuštění serveru ( `PORT=20128` , API a dashboard na stejném portu)
-`omniroute --port 3000` | Nastavte kanonický/API port na 3000
-`omniroute --mcp` | Spuštění MCP serveru (transport stdio)
-`omniroute --no-open` | Neotevírat prohlížeč automaticky
-`omniroute --help` | Zobrazit nápovědu
+| Příkaz                  | Popis                                                               |
+| ----------------------- | ------------------------------------------------------------------- |
+| `omniroute`             | Spuštění serveru ( `PORT=20128` , API a dashboard na stejném portu) |
+| `omniroute --port 3000` | Nastavte kanonický/API port na 3000                                 |
+| `omniroute --mcp`       | Spuštění MCP serveru (transport stdio)                              |
+| `omniroute --no-open`   | Neotevírat prohlížeč automaticky                                    |
+| `omniroute --help`      | Zobrazit nápovědu                                                   |
 
 Volitelný režim s rozděleným portem:
 
@@ -847,10 +753,10 @@ docker compose --profile base up -d
 docker compose --profile cli up -d
 ```
 
-Obraz | Štítek | Velikost | Popis
---- | --- | --- | ---
-`diegosouzapw/omniroute` | `latest` | ~250 MB | Nejnovější stabilní verze
-`diegosouzapw/omniroute` | `1.0.3` | ~250 MB | Aktuální verze
+| Obraz                    | Štítek   | Velikost | Popis                     |
+| ------------------------ | -------- | -------- | ------------------------- |
+| `diegosouzapw/omniroute` | `latest` | ~250 MB  | Nejnovější stabilní verze |
+| `diegosouzapw/omniroute` | `1.0.3`  | ~250 MB  | Aktuální verze            |
 
 ---
 
@@ -893,41 +799,47 @@ Po minimalizaci se OmniRoute nachází v systémové liště a nabízí rychlé
 
 ## 💰 Přehled cen
 
-Úroveň | Poskytovatel | Náklady | Obnovení kvóty | Nejlepší pro
---- | --- | --- | --- | ---
-**💳 PŘEDPLATNÉ** | Claude Code (profesionál) | 20 dolarů měsíčně | 5 hodin + týdně | Již přihlášen/a k odběru
- | Kodex (Plus/Pro) | 20–200 USD/měsíc | 5 hodin + týdně | Uživatelé OpenAI
- | Rozhraní příkazového řádku Gemini | **UVOLNIT** | 180 tisíc měsíčně + 1 tisíc denně | Každý!
- | GitHub Copilot | 10–19 USD/měsíc | Měsíční | Uživatelé GitHubu
-**🔑 KLÍČ API** | NVIDIA NIM | **ZDARMA** (vývoj navždy) | ~40 ot./min | 70+ otevřených modelů
- | Mozky | **ZDARMA** (1 milion tok/den) | 60 000 otáček za minutu / 30 ot./min | Nejrychlejší na světě
- | Groq | **ZDARMA** (30 ot./min.) | 14,4 tisíc otáček za minutu | Ultrarychlá lama/gema
- | DeepSeek V3.2 | 0,27/1,10 USD za 1 milion | Žádný | Nejlepší zdůvodnění ceny a kvality
- | xAI Grok-4 Rychlý | **0,20/0,50 USD za 1 milion** 🆕 | Žádný | Nejrychlejší + volání nástroje, ultranízké
- | xAI Grok-4 (standardní) | 0,20/1,50 USD za 1 milion 🆕 | Žádný | Vlajková loď Reasoning od xAI
- | Mistral | Zkušební verze zdarma + placené | Omezená sazba | Evropská umělá inteligence
- | OpenRouter | Platba za použití | Žádný | Více než 100 modelů agregováno.
-**💰 LEVNÉ** | GLM-5 (přes Z.AI) 🆕 | 0,5 USD/1 milion | Denně v 10:00 | Výstup 128 tisíc obrazových bodů, nejnovější vlajková loď
- | GLM-4.7 | 0,6 USD/1 milion | Denně v 10:00 | Záloha rozpočtu
- | MiniMax M2.5 🆕 | Vstup 0,3 USD/1 milion | 5hodinové válcování | Úvaha + agentní úkoly
- | MiniMax M2.1 | 0,2 USD/1 milion | 5hodinové válcování | Nejlevnější varianta
- | Kimi K2.5 (Moonshot API) 🆕 | Platba za použití | Žádný | Přímý přístup k Moonshot API
- | Kimi K2 | 9 dolarů měsíčně bez závazků | 10 milionů tokenů/měsíc | Předvídatelné náklady
-**🆓 ZDARMA** | iFlow | **0 dolarů** | Neomezený | 5 modelů neomezeně
- | Qwen | **0 dolarů** | Neomezený | 4 modely neomezeně
- | Kiro | **0 dolarů** | Neomezený | Claude Sonnet/Haiku (tvorce AWS)
+| Úroveň                            | Poskytovatel                     | Náklady                              | Obnovení kvóty                             | Nejlepší pro                                              |
+| --------------------------------- | -------------------------------- | ------------------------------------ | ------------------------------------------ | --------------------------------------------------------- |
+| **💳 PŘEDPLATNÉ**                 | Claude Code (profesionál)        | 20 dolarů měsíčně                    | 5 hodin + týdně                            | Již přihlášen/a k odběru                                  |
+| Kodex (Plus/Pro)                  | 20–200 USD/měsíc                 | 5 hodin + týdně                      | Uživatelé OpenAI                           |
+| Rozhraní příkazového řádku Gemini | **UVOLNIT**                      | 180 tisíc měsíčně + 1 tisíc denně    | Každý!                                     |
+| GitHub Copilot                    | 10–19 USD/měsíc                  | Měsíční                              | Uživatelé GitHubu                          |
+| **🔑 KLÍČ API**                   | NVIDIA NIM                       | **ZDARMA** (vývoj navždy)            | ~40 ot./min                                | 70+ otevřených modelů                                     |
+| Mozky                             | **ZDARMA** (1 milion tok/den)    | 60 000 otáček za minutu / 30 ot./min | Nejrychlejší na světě                      |
+| Groq                              | **ZDARMA** (30 ot./min.)         | 14,4 tisíc otáček za minutu          | Ultrarychlá lama/gema                      |
+| DeepSeek V3.2                     | 0,27/1,10 USD za 1 milion        | Žádný                                | Nejlepší zdůvodnění ceny a kvality         |
+| xAI Grok-4 Rychlý                 | **0,20/0,50 USD za 1 milion** 🆕 | Žádný                                | Nejrychlejší + volání nástroje, ultranízké |
+| xAI Grok-4 (standardní)           | 0,20/1,50 USD za 1 milion 🆕     | Žádný                                | Vlajková loď Reasoning od xAI              |
+| Mistral                           | Zkušební verze zdarma + placené  | Omezená sazba                        | Evropská umělá inteligence                 |
+| OpenRouter                        | Platba za použití                | Žádný                                | Více než 100 modelů agregováno.            |
+| **💰 LEVNÉ**                      | GLM-5 (přes Z.AI) 🆕             | 0,5 USD/1 milion                     | Denně v 10:00                              | Výstup 128 tisíc obrazových bodů, nejnovější vlajková loď |
+| GLM-4.7                           | 0,6 USD/1 milion                 | Denně v 10:00                        | Záloha rozpočtu                            |
+| MiniMax M2.5 🆕                   | Vstup 0,3 USD/1 milion           | 5hodinové válcování                  | Úvaha + agentní úkoly                      |
+| MiniMax M2.1                      | 0,2 USD/1 milion                 | 5hodinové válcování                  | Nejlevnější varianta                       |
+| Kimi K2.5 (Moonshot API) 🆕       | Platba za použití                | Žádný                                | Přímý přístup k Moonshot API               |
+| Kimi K2                           | 9 dolarů měsíčně bez závazků     | 10 milionů tokenů/měsíc              | Předvídatelné náklady                      |
+| **🆓 ZDARMA**                     | iFlow                            | **0 dolarů**                         | Neomezený                                  | 5 modelů neomezeně                                        |
+| Qwen                              | **0 dolarů**                     | Neomezený                            | 4 modely neomezeně                         |
+| Kiro                              | **0 dolarů**                     | Neomezený                            | Claude Sonnet/Haiku (tvorce AWS)           |
 
 > 🆕 **Přidány nové modely (březen 2026):** řada Grok-4 Fast za 0,20 USD/0,50 USD/M (benchmarkováno na 1143 ms – o 30 % rychlejší než Gemini 2.5 Flash), GLM-5 přes Z.AI s výstupem 128K, uvažování MiniMax M2.5, aktualizované ceny DeepSeek V3.2, Kimi K2.5 přes Moonshot Direct API.
 
 **💡 Kombinovaný balík za 0 $ — Kompletní bezplatná instalace:**
 
 ```
-Gemini CLI (180K/mo free)
-  → iFlow (unlimited: kimi-k2-thinking, qwen3-coder-plus, deepseek-r1)
-  → Kiro (Claude Sonnet 4.5 + Haiku — unlimited, via AWS Builder ID)
-  → Qwen (4 models — unlimited)
-  → Groq (14.4K req/day — ultra-fast)
-  → NVIDIA NIM (70+ models — 40 RPM forever)
+# 🆓 Ultimate Free Stack 2026 — 11 Providers, $0 Forever
+Kiro (kr/)             → Claude Sonnet/Haiku UNLIMITED
+iFlow (if/)            → kimi-k2-thinking, qwen3-coder-plus, deepseek-r1 UNLIMITED
+LongCat Lite (lc/)     → LongCat-Flash-Lite — 50M tokens/day 🔥
+Pollinations (pol/)    → GPT-5, Claude, DeepSeek, Llama 4 — no key needed
+Qwen (qw/)             → qwen3-coder-plus, qwen3-coder-flash, qwen3-coder-next UNLIMITED
+Gemini (gemini/)       → Gemini 2.5 Flash — 1,500 req/day free API key
+Cloudflare AI (cf/)    → Llama 70B, Gemma 3, Mistral — 10K Neurons/day
+Scaleway (scw/)        → Qwen3 235B, Llama 70B — 1M free tokens (EU)
+Groq (groq/)           → Llama/Gemma ultra-fast — 14.4K req/day
+NVIDIA NIM (nvidia/)   → 70+ open models — 40 RPM forever
+Cerebras (cerebras/)   → Llama/Qwen world-fastest — 1M tok/day
 ```
 
 **Nulové náklady. Nikdy nepřestávejte s kódováním.** Nakonfigurujte si to jako jednu kombinaci OmniRoute a všechny záložní režimy se provede automaticky – žádné ruční přepínání.
@@ -942,59 +854,59 @@ Gemini CLI (180K/mo free)
 
 ### 🔵 CLAUDE MODELS (přes Kiro — AWS Builder ID)
 
-Model | Předpona | Omezit | Limit rychlosti
---- | --- | --- | ---
-`claude-sonnet-4.5` | `kr/` | **Neomezený** | Žádný hlášený denní limit
-`claude-haiku-4.5` | `kr/` | **Neomezený** | Žádný hlášený denní limit
-`claude-opus-4.6` | `kr/` | **Neomezený** | Nejnovější opus od Kira
+| Model               | Předpona | Omezit        | Limit rychlosti           |
+| ------------------- | -------- | ------------- | ------------------------- |
+| `claude-sonnet-4.5` | `kr/`    | **Neomezený** | Žádný hlášený denní limit |
+| `claude-haiku-4.5`  | `kr/`    | **Neomezený** | Žádný hlášený denní limit |
+| `claude-opus-4.6`   | `kr/`    | **Neomezený** | Nejnovější opus od Kira   |
 
 ### 🟢 MODELY IFLOW (Bezplatné OAuth — bez nutnosti platit kreditní kartou)
 
-Model | Předpona | Omezit | Limit rychlosti
---- | --- | --- | ---
-`kimi-k2-thinking` | `if/` | **Neomezený** | Žádný hlášený strop
-`qwen3-coder-plus` | `if/` | **Neomezený** | Žádný hlášený strop
-`deepseek-r1` | `if/` | **Neomezený** | Žádný hlášený strop
-`minimax-m2.1` | `if/` | **Neomezený** | Žádný hlášený strop
-`kimi-k2` | `if/` | **Neomezený** | Žádný hlášený strop
+| Model              | Předpona | Omezit        | Limit rychlosti     |
+| ------------------ | -------- | ------------- | ------------------- |
+| `kimi-k2-thinking` | `if/`    | **Neomezený** | Žádný hlášený strop |
+| `qwen3-coder-plus` | `if/`    | **Neomezený** | Žádný hlášený strop |
+| `deepseek-r1`      | `if/`    | **Neomezený** | Žádný hlášený strop |
+| `minimax-m2.1`     | `if/`    | **Neomezený** | Žádný hlášený strop |
+| `kimi-k2`          | `if/`    | **Neomezený** | Žádný hlášený strop |
 
 ### 🟡 MODELY QWEN (Ověření kódu zařízení)
 
-Model | Předpona | Omezit | Limit rychlosti
---- | --- | --- | ---
-`qwen3-coder-plus` | `qw/` | **Neomezený** | Žádný hlášený strop
-`qwen3-coder-flash` | `qw/` | **Neomezený** | Žádný hlášený strop
-`qwen3-coder-next` | `qw/` | **Neomezený** | Žádný hlášený strop
-`vision-model` | `qw/` | **Neomezený** | Multimodální (obrázky)
+| Model               | Předpona | Omezit        | Limit rychlosti        |
+| ------------------- | -------- | ------------- | ---------------------- |
+| `qwen3-coder-plus`  | `qw/`    | **Neomezený** | Žádný hlášený strop    |
+| `qwen3-coder-flash` | `qw/`    | **Neomezený** | Žádný hlášený strop    |
+| `qwen3-coder-next`  | `qw/`    | **Neomezený** | Žádný hlášený strop    |
+| `vision-model`      | `qw/`    | **Neomezený** | Multimodální (obrázky) |
 
 ### 🟣 Rozhraní GEMINI CLI (Google OAuth)
 
-Model | Předpona | Omezit | Limit rychlosti
---- | --- | --- | ---
-`gemini-3-flash-preview` | `gc/` | **180 tisíc tok/měsíc** + 1 tisíc/den | Měsíční reset
-`gemini-2.5-pro` | `gc/` | 180 tisíc měsíčně (sdílený bazén) | Vysoká kvalita
+| Model                    | Předpona | Omezit                                | Limit rychlosti |
+| ------------------------ | -------- | ------------------------------------- | --------------- |
+| `gemini-3-flash-preview` | `gc/`    | **180 tisíc tok/měsíc** + 1 tisíc/den | Měsíční reset   |
+| `gemini-2.5-pro`         | `gc/`    | 180 tisíc měsíčně (sdílený bazén)     | Vysoká kvalita  |
 
 ### ⚫ NVIDIA NIM (Bezplatný klíč API — build.nvidia.com)
 
-Úroveň | Denní limit | Limit rychlosti | Poznámky
---- | --- | --- | ---
-Zdarma (vývojář) | Žádný limit tokenů | **~40 ot./min** | Více než 70 modelů; přechod na čisté limity sazeb v polovině roku 2025
+| Úroveň           | Denní limit        | Limit rychlosti | Poznámky                                                               |
+| ---------------- | ------------------ | --------------- | ---------------------------------------------------------------------- |
+| Zdarma (vývojář) | Žádný limit tokenů | **~40 ot./min** | Více než 70 modelů; přechod na čisté limity sazeb v polovině roku 2025 |
 
 Oblíbené bezplatné modely: `moonshotai/kimi-k2.5` (Kimi K2.5), `z-ai/glm4.7` (GLM 4.7), `deepseek-ai/deepseek-v3.2` (DeepSeek V3.2), `nvidia/llama-3.3-70b-instruct` , `deepseek/deepseek-r1`
 
 ### ⚪ CEREBRAS (Bezplatný klíč API — inference.cerebras.ai)
 
-Úroveň | Denní limit | Limit rychlosti | Poznámky
---- | --- | --- | ---
-Uvolnit | **1 milion tokenů/den** | 60 000 otáček za minutu / 30 ot./min | Nejrychlejší inference LLM na světě; denně se resetuje
+| Úroveň  | Denní limit             | Limit rychlosti                      | Poznámky                                               |
+| ------- | ----------------------- | ------------------------------------ | ------------------------------------------------------ |
+| Uvolnit | **1 milion tokenů/den** | 60 000 otáček za minutu / 30 ot./min | Nejrychlejší inference LLM na světě; denně se resetuje |
 
 Dostupné zdarma: `llama-3.3-70b` , `llama-3.1-8b` , `deepseek-r1-distill-llama-70b`
 
 ### 🔴 GROQ (Bezplatný API klíč — console.groq.com)
 
-Úroveň | Denní limit | Limit rychlosti | Poznámky
---- | --- | --- | ---
-Uvolnit | **14,4 tisíc otáček za minutu** | 30 ot./min na model | Žádná kreditní karta; limit 429, neúčtováno
+| Úroveň  | Denní limit                     | Limit rychlosti     | Poznámky                                    |
+| ------- | ------------------------------- | ------------------- | ------------------------------------------- |
+| Uvolnit | **14,4 tisíc otáček za minutu** | 30 ot./min na model | Žádná kreditní karta; limit 429, neúčtováno |
 
 K dispozici zdarma: `llama-3.3-70b-versatile` , `gemma2-9b-it` , `mixtral-8x7b` , `whisper-large-v3`
 
@@ -1016,11 +928,11 @@ K dispozici zdarma: `llama-3.3-70b-versatile` , `gemma2-9b-it` , `mixtral-8x7b`
 
 > Přepisujte libovolné audio/video za **0 $** – Deepgram leady za 200 $ zdarma, AssemblyAI za 50 $ jako záložní nástroj, Groq Whisper jako neomezená nouzová záloha.
 
-Poskytovatel | Bezplatné kredity | Nejlepší model | Limit rychlosti
---- | --- | --- | ---
-🟢 **Deepgram** | **200 dolarů zdarma** (registrace) | `nova-3` — nejvyšší přesnost, více než 30 jazyků | Žádný limit RPM pro kredity zdarma
-🔵 **AssemblyAI** | **50 dolarů zdarma** (registrace) | `universal-3-pro` — kapitoly, sentiment, osobní údaje | Žádný limit RPM pro kredity zdarma
-🔴 **Groq** | **Navždy zdarma** | `whisper-large-v3` — OpenAI Šepot | 30 ot./min (omezená rychlost)
+| Poskytovatel      | Bezplatné kredity                  | Nejlepší model                                        | Limit rychlosti                    |
+| ----------------- | ---------------------------------- | ----------------------------------------------------- | ---------------------------------- |
+| 🟢 **Deepgram**   | **200 dolarů zdarma** (registrace) | `nova-3` — nejvyšší přesnost, více než 30 jazyků      | Žádný limit RPM pro kredity zdarma |
+| 🔵 **AssemblyAI** | **50 dolarů zdarma** (registrace)  | `universal-3-pro` — kapitoly, sentiment, osobní údaje | Žádný limit RPM pro kredity zdarma |
+| 🔴 **Groq**       | **Navždy zdarma**                  | `whisper-large-v3` — OpenAI Šepot                     | 30 ot./min (omezená rychlost)      |
 
 **Navrhovaná kombinace v `/dashboard/combos` :**
 
@@ -1041,118 +953,118 @@ OmniRoute v2.0 je navržen jako operační platforma, nikoli pouze jako proxy pr
 
 ### 🆕 Nové — Vylepšení inspirovaná ClawRouterem (březen 2026)
 
-Funkce | Co to dělá
---- | ---
-⚡ **Grok-4 Rychlá rodina** | Modely xAI za 0,20 USD/0,50 USD/M – v benchmarku 1143 ms (o 30 % rychlejší než Gemini 2.5 Flash)
-🧠 **GLM-5 přes Z.AI** | 128 tisíc výstupních dat, 0,5 USD/1 milion USD – nejnovější vlajková loď rodiny GLM
-🔮 **MiniMax M2.5** | Úvaha + agentní úkoly za 0,30 USD/1 milion – významný upgrade oproti M2.1
-🎯 **Příznak volání nástroje pro každý model** | `toolCalling: true/false` v registru — AutoCombo přeskakuje modely, které nepodporují nástroje.
-🌍 **Detekce vícejazyčného záměru** | Klíčová slova PT/ZH/ES/AR v bodování AutoCombo – lepší výběr modelu pro neanglický obsah
-📊 **Záložní metody řízené benchmarkem** | Skutečná latence p95 z živých požadavků poskytuje kombinované skóre – AutoCombo se učí ze skutečných dat
-🔁 **Požádat o deduplikaci** | Okno pro deduplikaci na základě hashování obsahu – bezpečné pro více agentů, zabraňuje duplicitním platbám
-🔌 **Strategie pro zásuvné routery** | Rozšiřitelné rozhraní `RouterStrategy` – přidejte si vlastní logiku směrování jako pluginy
+| Funkce                                         | Co to dělá                                                                                                 |
+| ---------------------------------------------- | ---------------------------------------------------------------------------------------------------------- |
+| ⚡ **Grok-4 Rychlá rodina**                    | Modely xAI za 0,20 USD/0,50 USD/M – v benchmarku 1143 ms (o 30 % rychlejší než Gemini 2.5 Flash)           |
+| 🧠 **GLM-5 přes Z.AI**                         | 128 tisíc výstupních dat, 0,5 USD/1 milion USD – nejnovější vlajková loď rodiny GLM                        |
+| 🔮 **MiniMax M2.5**                            | Úvaha + agentní úkoly za 0,30 USD/1 milion – významný upgrade oproti M2.1                                  |
+| 🎯 **Příznak volání nástroje pro každý model** | `toolCalling: true/false` v registru — AutoCombo přeskakuje modely, které nepodporují nástroje.            |
+| 🌍 **Detekce vícejazyčného záměru**            | Klíčová slova PT/ZH/ES/AR v bodování AutoCombo – lepší výběr modelu pro neanglický obsah                   |
+| 📊 **Záložní metody řízené benchmarkem**       | Skutečná latence p95 z živých požadavků poskytuje kombinované skóre – AutoCombo se učí ze skutečných dat   |
+| 🔁 **Požádat o deduplikaci**                   | Okno pro deduplikaci na základě hashování obsahu – bezpečné pro více agentů, zabraňuje duplicitním platbám |
+| 🔌 **Strategie pro zásuvné routery**           | Rozšiřitelné rozhraní `RouterStrategy` – přidejte si vlastní logiku směrování jako pluginy                 |
 
 ### 🚀 Předchozí verze v2.0.9+ — Hřiště, otisky prstů v CLI a ACP
 
-Funkce | Co to dělá
---- | ---
-🎮 **Modelové hřiště** | Stránka řídicího panelu pro přímé testování libovolného modelu – selektory poskytovatele/modelu/koncového bodu, editor Monaco, streamování, přerušení, načasování
-🔏 **Porovnávání otisků prstů v CLI** | Řazení hlaviček/těl serveru podle poskytovatele tak, aby odpovídalo nativním podpisům CLI – přepínání pro jednotlivé poskytovatele v Nastavení &gt; Zabezpečení. **Vaše IP adresa proxy serveru je zachována.**
-🤝 **Podpora ACP (Agent Client Protocol)** | Vyhledávání agentů CLI (Codex, Claude, Goose, Gemini CLI, OpenClaw + 9 dalších), generátor procesů, koncový bod `/api/acp/agents`
-🤖 **Řídicí panel agentů ACP** | Ladění › Stránka Agenti — mřížka 14 agentů se stavem instalace, verzí a formulářem pro vlastní agenta pro libovolný nástroj CLI. Uživatelé **OpenCode** získají tlačítko „Stáhnout opencode.json“, které automaticky vygeneruje konfiguraci připravenou k použití se všemi dostupnými modely.
-🔧 **Směrování `apiFormat` pro vlastní model** | Vlastní modely s `apiFormat: "responses"` nyní správně směrují do překladače Responses API.
-🏢 **Izolace pracovního prostoru Codexu** | Více pracovních prostorů Codexu na jeden e-mail – OAuth správně odděluje připojení podle ID pracovního prostoru
-🔄 **Automatická aktualizace elektronů** | Desktopová aplikace kontroluje aktualizace + automaticky se instaluje po restartu
+| Funkce                                         | Co to dělá                                                                                                                                                                                                                                                                                    |
+| ---------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🎮 **Modelové hřiště**                         | Stránka řídicího panelu pro přímé testování libovolného modelu – selektory poskytovatele/modelu/koncového bodu, editor Monaco, streamování, přerušení, načasování                                                                                                                             |
+| 🔏 **Porovnávání otisků prstů v CLI**          | Řazení hlaviček/těl serveru podle poskytovatele tak, aby odpovídalo nativním podpisům CLI – přepínání pro jednotlivé poskytovatele v Nastavení &gt; Zabezpečení. **Vaše IP adresa proxy serveru je zachována.**                                                                               |
+| 🤝 **Podpora ACP (Agent Client Protocol)**     | Vyhledávání agentů CLI (Codex, Claude, Goose, Gemini CLI, OpenClaw + 9 dalších), generátor procesů, koncový bod `/api/acp/agents`                                                                                                                                                             |
+| 🤖 **Řídicí panel agentů ACP**                 | Ladění › Stránka Agenti — mřížka 14 agentů se stavem instalace, verzí a formulářem pro vlastní agenta pro libovolný nástroj CLI. Uživatelé **OpenCode** získají tlačítko „Stáhnout opencode.json“, které automaticky vygeneruje konfiguraci připravenou k použití se všemi dostupnými modely. |
+| 🔧 **Směrování `apiFormat` pro vlastní model** | Vlastní modely s `apiFormat: "responses"` nyní správně směrují do překladače Responses API.                                                                                                                                                                                                   |
+| 🏢 **Izolace pracovního prostoru Codexu**      | Více pracovních prostorů Codexu na jeden e-mail – OAuth správně odděluje připojení podle ID pracovního prostoru                                                                                                                                                                               |
+| 🔄 **Automatická aktualizace elektronů**       | Desktopová aplikace kontroluje aktualizace + automaticky se instaluje po restartu                                                                                                                                                                                                             |
 
 ### 🤖 Operace s agenty a protokoly (v2.0)
 
-Funkce | Co to dělá
---- | ---
-🔧 **MCP Server (16 nástrojů)** | Nástroje IDE/agent prostřednictvím 3 transportů: stdio, SSE ( `/api/mcp/sse` ), Streamovatelný HTTP ( `/api/mcp/stream` )
-🤝 **A2A server (JSON-RPC + SSE)** | Spouštění úloh mezi agenty se synchronizací a streamováním
-🧭 **Konsolidovaná stránka koncových bodů** | Stránka pro správu s kartami Endpoint Proxy, MCP, A2A a API Endpoints
-🎚️ **Přepínače pro povolení/zakázání služby** | Přepínače ZAP/VYP pro MCP a A2A s trvalým nastavením (výchozí: VYP)
-🛰️ **Srdeční tep za běhu MCP** | Skutečný stav procesu (pid, doba provozuschopnosti, stáří heartbeatu, transport, režim rozsahu)
-📋 **Auditní záznam MCP** | Filtrovatelné protokoly auditu s hodnocením úspěchu/neúspěchu a klíčovým přiřazením
-🔐 **Vynucování rozsahu MCP** | 9 podrobných oprávnění pro řízený přístup k nástrojům
-📡 **Správa životního cyklu úkolů A2A** | Seznam/filtrování úloh, kontrola událostí/artefaktů, zrušení spuštěných úloh
-📋 **Objevení karty agenta** | `/.well-known/agent.json` pro automatické vyhledávání klientů
-🧪 **Testovací postroj Protocol E2E** | Skutečné MCP SDK + toky klientů A2A v `test:protocols:e2e`
-⚙️ **Provozní kontroly** | Kombinace přepínačů, použití profilů odolnosti, resetování jističů z jednoho ovládacího panelu
+| Funkce                                        | Co to dělá                                                                                                                |
+| --------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------- |
+| 🔧 **MCP Server (16 nástrojů)**               | Nástroje IDE/agent prostřednictvím 3 transportů: stdio, SSE ( `/api/mcp/sse` ), Streamovatelný HTTP ( `/api/mcp/stream` ) |
+| 🤝 **A2A server (JSON-RPC + SSE)**            | Spouštění úloh mezi agenty se synchronizací a streamováním                                                                |
+| 🧭 **Konsolidovaná stránka koncových bodů**   | Stránka pro správu s kartami Endpoint Proxy, MCP, A2A a API Endpoints                                                     |
+| 🎚️ **Přepínače pro povolení/zakázání služby** | Přepínače ZAP/VYP pro MCP a A2A s trvalým nastavením (výchozí: VYP)                                                       |
+| 🛰️ **Srdeční tep za běhu MCP**                | Skutečný stav procesu (pid, doba provozuschopnosti, stáří heartbeatu, transport, režim rozsahu)                           |
+| 📋 **Auditní záznam MCP**                     | Filtrovatelné protokoly auditu s hodnocením úspěchu/neúspěchu a klíčovým přiřazením                                       |
+| 🔐 **Vynucování rozsahu MCP**                 | 9 podrobných oprávnění pro řízený přístup k nástrojům                                                                     |
+| 📡 **Správa životního cyklu úkolů A2A**       | Seznam/filtrování úloh, kontrola událostí/artefaktů, zrušení spuštěných úloh                                              |
+| 📋 **Objevení karty agenta**                  | `/.well-known/agent.json` pro automatické vyhledávání klientů                                                             |
+| 🧪 **Testovací postroj Protocol E2E**         | Skutečné MCP SDK + toky klientů A2A v `test:protocols:e2e`                                                                |
+| ⚙️ **Provozní kontroly**                      | Kombinace přepínačů, použití profilů odolnosti, resetování jističů z jednoho ovládacího panelu                            |
 
 ### 🧠 Směrování a inteligence
 
-Funkce | Co to dělá
---- | ---
-🎯 **Inteligentní čtyřúrovňový záložní systém** | Automatická trasa: Předplatné → API klíč → Levné → Zdarma
-📊 **Sledování kvót v reálném čase** | Počet tokenů v reálném čase + odpočet resetování pro každého poskytovatele
-🔄 **Překlad formátu** | OpenAI ↔ Claude ↔ Gemini ↔ Odpovědi s konverzemi bezpečnými pro schéma
-👥 **Podpora více účtů** | Více účtů na poskytovatele s inteligentním výběrem
-🔄 **Automatická aktualizace tokenů** | Tokeny OAuth se automaticky obnovují při opakovaném pokusu.
-🎨 **Vlastní kombinace** | 6 vyvažovacích strategií + řízení záložního řetězce
-🌐 **Směrovač se zástupnými znaky** | dynamické směrování `provider/*`
-🧠 **Přemýšlení o rozpočtových kontrolách** | Limity pro průchozí, automatické, vlastní a adaptivní uvažování
-🔀 **Aliasy modelů** | Vestavěné + vlastní aliasování modelů a bezpečnost migrace
-⚡ **Degradace pozadí** | Směrujte úlohy na pozadí s nízkou prioritou na levnější modely
-🧪 **Chytré směrování s ohledem na úkoly** | Automatický výběr modelu podle typu obsahu (kódování/vize/analýza/sumarizace)
-💬 **Vstřikování do systému** | Globální kontroly chování uplatňované konzistentně
-📄 **Kompatibilita API pro odpovědi** | Plná podpora `/v1/responses` pro Codex a pokročilé agentické pracovní postupy
+| Funkce                                          | Co to dělá                                                                    |
+| ----------------------------------------------- | ----------------------------------------------------------------------------- |
+| 🎯 **Inteligentní čtyřúrovňový záložní systém** | Automatická trasa: Předplatné → API klíč → Levné → Zdarma                     |
+| 📊 **Sledování kvót v reálném čase**            | Počet tokenů v reálném čase + odpočet resetování pro každého poskytovatele    |
+| 🔄 **Překlad formátu**                          | OpenAI ↔ Claude ↔ Gemini ↔ Odpovědi s konverzemi bezpečnými pro schéma        |
+| 👥 **Podpora více účtů**                        | Více účtů na poskytovatele s inteligentním výběrem                            |
+| 🔄 **Automatická aktualizace tokenů**           | Tokeny OAuth se automaticky obnovují při opakovaném pokusu.                   |
+| 🎨 **Vlastní kombinace**                        | 6 vyvažovacích strategií + řízení záložního řetězce                           |
+| 🌐 **Směrovač se zástupnými znaky**             | dynamické směrování `provider/*`                                              |
+| 🧠 **Přemýšlení o rozpočtových kontrolách**     | Limity pro průchozí, automatické, vlastní a adaptivní uvažování               |
+| 🔀 **Aliasy modelů**                            | Vestavěné + vlastní aliasování modelů a bezpečnost migrace                    |
+| ⚡ **Degradace pozadí**                         | Směrujte úlohy na pozadí s nízkou prioritou na levnější modely                |
+| 🧪 **Chytré směrování s ohledem na úkoly**      | Automatický výběr modelu podle typu obsahu (kódování/vize/analýza/sumarizace) |
+| 💬 **Vstřikování do systému**                   | Globální kontroly chování uplatňované konzistentně                            |
+| 📄 **Kompatibilita API pro odpovědi**           | Plná podpora `/v1/responses` pro Codex a pokročilé agentické pracovní postupy |
 
 ### 🎵 Multimodální API
 
-Funkce | Co to dělá
---- | ---
-🖼️ **Generování obrázků** | `/v1/images/generations` s cloudovým a lokálním backendem
-📐 **Vložení** | `/v1/embeddings` pro vyhledávání a RAG pipelines
-🎤 **Přepis zvuku** | `/v1/audio/transcriptions` (Whisper a další poskytovatelé)
-🔊 **Převod textu na řeč** | `/v1/audio/speech` (více enginů/poskytovatelů)
-🎬 **Generování videa** | `/v1/videos/generations` (pracovní postupy ComfyUI + SD WebUI)
-🎵 **Hudební generace** | `/v1/music/generations` (pracovní postupy ComfyUI)
-🛡️ **Moderování** | Bezpečnostní kontroly `/v1/moderations`
-🔀 **Změna pořadí** | `/v1/rerank` pro hodnocení relevance
-🔍 **Vyhledávání na webu** 🆕 | `/v1/search` — 5 poskytovatelů (Serper, Brave, Perplexity, Exa, Tavily), více než 6 500 zdarma/měsíc, automatické přepnutí na záložní systém, mezipaměť
+| Funkce                        | Co to dělá                                                                                                                                              |
+| ----------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Generování obrázků**     | `/v1/images/generations` s cloudovým a lokálním backendem                                                                                               |
+| 📐 **Vložení**                | `/v1/embeddings` pro vyhledávání a RAG pipelines                                                                                                        |
+| 🎤 **Přepis zvuku**           | `/v1/audio/transcriptions` (Whisper a další poskytovatelé)                                                                                              |
+| 🔊 **Převod textu na řeč**    | `/v1/audio/speech` (více enginů/poskytovatelů)                                                                                                          |
+| 🎬 **Generování videa**       | `/v1/videos/generations` (pracovní postupy ComfyUI + SD WebUI)                                                                                          |
+| 🎵 **Hudební generace**       | `/v1/music/generations` (pracovní postupy ComfyUI)                                                                                                      |
+| 🛡️ **Moderování**             | Bezpečnostní kontroly `/v1/moderations`                                                                                                                 |
+| 🔀 **Změna pořadí**           | `/v1/rerank` pro hodnocení relevance                                                                                                                    |
+| 🔍 **Vyhledávání na webu** 🆕 | `/v1/search` — 5 poskytovatelů (Serper, Brave, Perplexity, Exa, Tavily), více než 6 500 zdarma/měsíc, automatické přepnutí na záložní systém, mezipaměť |
 
 ### 🛡️ Odolnost, bezpečnost a správa věcí veřejných
 
-Funkce | Co to dělá
---- | ---
-🔌 **Jističe** | Vypnutí/obnovení pro každý model s ovládáním prahových hodnot
-🎯 **Modely s ohledem na koncové body** | Vlastní modely deklarují podporované koncové body + formát API
-🛡️ **Stádo proti hromům** | Ochrana mutexu a semaforu při událostech opakování/rychlosti
-🧠 **Sémantická + podpisová mezipaměť** | Snížení nákladů/latence díky dvěma vrstvám mezipaměti
-⚡ **Žádost o idempotenci** | Okno ochrany proti duplikacím
-🔒 **Falšování otisků prstů pomocí TLS** | Otisk TLS podobný prohlížeči – **snižuje detekci botů a nahlašování účtů**
-🔏 **Porovnávání otisků prstů v CLI** | Shoduje se s nativními podpisy požadavků CLI – **snižuje riziko zablokování a zároveň zachovává IP adresu proxy**
-🌐 **Filtrování IP adres** | Ovládání seznamu povolených/blokovaných položek pro odhalená nasazení
-📊 **Upravitelné limity rychlosti** | Konfigurovatelné globální/na úrovni poskytovatele limity s perzistencí
-🔑 **Správa klíčů API a stanovení rozsahu** | Bezpečné vydávání/rotace klíčů a kontroly modelu/poskytovatele
-🛡️ **Chráněné `/models`** | Volitelné ověřování a skrytí poskytovatele pro katalog modelů
+| Funkce                                      | Co to dělá                                                                                                        |
+| ------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- |
+| 🔌 **Jističe**                              | Vypnutí/obnovení pro každý model s ovládáním prahových hodnot                                                     |
+| 🎯 **Modely s ohledem na koncové body**     | Vlastní modely deklarují podporované koncové body + formát API                                                    |
+| 🛡️ **Stádo proti hromům**                   | Ochrana mutexu a semaforu při událostech opakování/rychlosti                                                      |
+| 🧠 **Sémantická + podpisová mezipaměť**     | Snížení nákladů/latence díky dvěma vrstvám mezipaměti                                                             |
+| ⚡ **Žádost o idempotenci**                 | Okno ochrany proti duplikacím                                                                                     |
+| 🔒 **Falšování otisků prstů pomocí TLS**    | Otisk TLS podobný prohlížeči – **snižuje detekci botů a nahlašování účtů**                                        |
+| 🔏 **Porovnávání otisků prstů v CLI**       | Shoduje se s nativními podpisy požadavků CLI – **snižuje riziko zablokování a zároveň zachovává IP adresu proxy** |
+| 🌐 **Filtrování IP adres**                  | Ovládání seznamu povolených/blokovaných položek pro odhalená nasazení                                             |
+| 📊 **Upravitelné limity rychlosti**         | Konfigurovatelné globální/na úrovni poskytovatele limity s perzistencí                                            |
+| 🔑 **Správa klíčů API a stanovení rozsahu** | Bezpečné vydávání/rotace klíčů a kontroly modelu/poskytovatele                                                    |
+| 🛡️ **Chráněné `/models`**                   | Volitelné ověřování a skrytí poskytovatele pro katalog modelů                                                     |
 
 ### 📊 Pozorovatelnost a analytika
 
-Funkce | Co to dělá
---- | ---
-📝 **Žádost + protokolování proxy** | Úplné protokolování požadavků/odpovědí a proxy
-📋 **Sjednocený panel protokolů** | Zobrazení požadavků, proxy, auditu a konzole na jedné stránce
-🔍 **Vyžádat si telemetrii** | Latence p50/p95/p99 a trasování požadavků
-🏥 **Panel zdraví** | Doba provozuschopnosti, stavy jističů, uzamčení, statistiky mezipaměti
-💰 **Sledování nákladů** | Kontrola rozpočtu a přehled o cenách pro jednotlivé modely
-📈 **Analytické vizualizace** | Přehledy využití modelů/poskytovatelů a zobrazení trendů
-🧪 **Rámec hodnocení** | Testování zlaté sady s konfigurovatelnými strategiemi shody
+| Funkce                              | Co to dělá                                                             |
+| ----------------------------------- | ---------------------------------------------------------------------- |
+| 📝 **Žádost + protokolování proxy** | Úplné protokolování požadavků/odpovědí a proxy                         |
+| 📋 **Sjednocený panel protokolů**   | Zobrazení požadavků, proxy, auditu a konzole na jedné stránce          |
+| 🔍 **Vyžádat si telemetrii**        | Latence p50/p95/p99 a trasování požadavků                              |
+| 🏥 **Panel zdraví**                 | Doba provozuschopnosti, stavy jističů, uzamčení, statistiky mezipaměti |
+| 💰 **Sledování nákladů**            | Kontrola rozpočtu a přehled o cenách pro jednotlivé modely             |
+| 📈 **Analytické vizualizace**       | Přehledy využití modelů/poskytovatelů a zobrazení trendů               |
+| 🧪 **Rámec hodnocení**              | Testování zlaté sady s konfigurovatelnými strategiemi shody            |
 
 ### ☁️ Nasazení a platforma
 
-Funkce | Co to dělá
---- | ---
-🌐 **Nasazení kdekoli** | Localhost, VPS, Docker, cloudová prostředí
-💾 **Synchronizace s cloudem** | Synchronizace konfigurace přes cloud worker
-🔄 **Zálohování/Obnovení** | Toky exportu/importu a obnovy po havárii
-🧙 **Průvodce nástupem** | Průvodce prvním spuštěním
-🔧 **Panel nástrojů CLI** | Nastavení oblíbených kódovacích nástrojů jedním kliknutím
-🎮 **Modelové hřiště** | Otestujte libovolného poskytovatele/model/koncový bod z řídicího panelu
-🔏 **Přepínač otisků prstů v příkazovém řádku** | Porovnávání otisků prstů podle poskytovatele v Nastavení &gt; Zabezpečení
-🌐 **i18n (30 jazyků)** | Plná jazyková podpora dashboardu a dokumentace s psaním zprava doleva
-📂 **Adresář vlastních dat** | Přepsání `DATA_DIR` pro umístění úložiště
+| Funkce                                          | Co to dělá                                                                |
+| ----------------------------------------------- | ------------------------------------------------------------------------- |
+| 🌐 **Nasazení kdekoli**                         | Localhost, VPS, Docker, cloudová prostředí                                |
+| 💾 **Synchronizace s cloudem**                  | Synchronizace konfigurace přes cloud worker                               |
+| 🔄 **Zálohování/Obnovení**                      | Toky exportu/importu a obnovy po havárii                                  |
+| 🧙 **Průvodce nástupem**                        | Průvodce prvním spuštěním                                                 |
+| 🔧 **Panel nástrojů CLI**                       | Nastavení oblíbených kódovacích nástrojů jedním kliknutím                 |
+| 🎮 **Modelové hřiště**                          | Otestujte libovolného poskytovatele/model/koncový bod z řídicího panelu   |
+| 🔏 **Přepínač otisků prstů v příkazovém řádku** | Porovnávání otisků prstů podle poskytovatele v Nastavení &gt; Zabezpečení |
+| 🌐 **i18n (30 jazyků)**                         | Plná jazyková podpora dashboardu a dokumentace s psaním zprava doleva     |
+| 📂 **Adresář vlastních dat**                    | Přepsání `DATA_DIR` pro umístění úložiště                                 |
 
 ### Hluboký pohled na funkce
 
@@ -1203,12 +1115,12 @@ Předinstalovaná sada „OmniRoute Golden Set“ obsahuje testovací případy
 
 ### Strategie hodnocení
 
-Strategie | Popis | Příklad
---- | --- | ---
-`exact` | Výstup se musí přesně shodovat | `"4"`
-`contains` | Výstup musí obsahovat podřetězec (bez rozlišení velkých a malých písmen) | `"Paris"`
-`regex` | Výstup musí odpovídat vzoru regulárních výrazů | `"1.*2.*3"`
-`custom` | Vlastní JS funkce vrací true/false | `(output) => output.length > 10`
+| Strategie  | Popis                                                                    | Příklad                          |
+| ---------- | ------------------------------------------------------------------------ | -------------------------------- |
+| `exact`    | Výstup se musí přesně shodovat                                           | `"4"`                            |
+| `contains` | Výstup musí obsahovat podřetězec (bez rozlišení velkých a malých písmen) | `"Paris"`                        |
+| `regex`    | Výstup musí odpovídat vzoru regulárních výrazů                           | `"1.*2.*3"`                      |
+| `custom`   | Vlastní JS funkce vrací true/false                                       | `(output) => output.length > 10` |
 
 ---
 
@@ -1240,9 +1152,6 @@ Užitečná API pro automatizaci:
 - `GET /api/mcp/audit`
 - `GET /api/mcp/audit/stats`
 
-
-
-
 <details>
 <summary><b>🤝 Nastavení A2A (Agent2Agent)</b></summary>
 </details>
@@ -1272,9 +1181,6 @@ Provozní uživatelské rozhraní:
 
 - `/dashboard/a2a` pro pozorovatelnost úloh/stavů/streamů a akce kouření
 
-
-
-
 <details>
 <summary><b>🧪 Komplexní validace protokolu</b></summary>
 </details>
@@ -1291,9 +1197,6 @@ Tím se ověřuje:
 - A2A objevování/odesílání/streamování/získávání/zrušení
 - Křížová kontrola dat v auditu MCP a API pro správu úloh A2A
 
-
-
-
 <details>
 <summary><b>💳 Poskytovatelé předplatného</b></summary>
 </details>
@@ -1369,9 +1272,6 @@ Models:
   gh/gemini-3-pro
 ```
 
-
-
-
 <details>
 <summary><b>🔑 Poskytovatelé klíčů API</b></summary>
 </details>
@@ -1381,7 +1281,7 @@ Models:
 1. Registrace: [build.nvidia.com](https://build.nvidia.com)
 2. Získejte zdarma klíč API (včetně 1000 inferenčních kreditů)
 3. Ovládací panel → Přidat poskytovatele → NVIDIA NIM:
-    - Klíč API: `nvapi-your-key`
+   - Klíč API: `nvapi-your-key`
 
 **Modely:** `nvidia/llama-3.3-70b-instruct` , `nvidia/mistral-7b-instruct` a více než 50 dalších
 
@@ -1413,9 +1313,6 @@ Models:
 
 **Modely:** Získejte přístup k více než 100 modelům od všech hlavních poskytovatelů prostřednictvím jediného klíče API.
 
-
-
-
 <details>
 <summary><b>💰 Levní poskytovatelé (záložní)</b></summary>
 </details>
@@ -1425,8 +1322,8 @@ Models:
 1. Registrace: [Zhipu AI](https://open.bigmodel.cn/)
 2. Získejte klíč API z kódovacího plánu
 3. Nástěnka → Přidat klíč API:
-    - Poskytovatel: `glm`
-    - Klíč API: `your-key`
+   - Poskytovatel: `glm`
+   - Klíč API: `your-key`
 
 **Použití:** `glm/glm-4.7`
 
@@ -1452,9 +1349,6 @@ Models:
 
 **Tip pro profesionály:** Fixních 9 $/měsíc za 10 milionů tokenů = efektivní náklady 0,90 $/1 milion!
 
-
-
-
 <details>
 <summary><b>🆓 BEZPLATNÍ poskytovatelé (nouzové zálohování)</b></summary>
 </details>
@@ -1498,9 +1392,6 @@ Models:
   kr/claude-haiku-4.5
 ```
 
-
-
-
 <details>
 <summary><b>🎨 Vytvořte kombinace</b></summary>
 </details>
@@ -1531,9 +1422,6 @@ Models:
 Cost: $0 forever!
 ```
 
-
-
-
 <details>
 <summary><b>🔧 Integrace s rozhraním příkazového řádku</b></summary>
 </details>
@@ -1637,9 +1525,6 @@ opencode
 
 > **Tip:** Do sekce `models` přidejte jakýkoli model dostupný ve vašem koncovém bodu OmniRoute `/v1/models` . Použijte formát `provider/model-id` z vašeho dashboardu OmniRoute.
 
-
-
-
 ---
 
 ## 🐛 Řešení problémů
@@ -1880,14 +1765,8 @@ Chcete-li získat přístup k kriterii pověření, můžete použít adresu **U
 
 > Toto řešení funguje na základě autorizačního kódu na adrese URL a nezávislého přesměrování přesměrování nebo jiného.
 
-
-
-
 ---
 
-
-
-
 ## 🛠️ Technologický stack
 
 <details>
@@ -1909,28 +1788,25 @@ Chcete-li získat přístup k kriterii pověření, můžete použít adresu **U
 - **Docker** : [hub.docker.com/r/diegosouzapw/omniroute](https://hub.docker.com/r/diegosouzapw/omniroute)
 - **Odolnost** : Jistič, exponenciální odstavení, ochrana proti hromům, falešné TLS, automatické kombinované samoopravování
 
-
-
-
 ---
 
 ## 📖 Dokumentace
 
-Dokument | Popis
---- | ---
-[Uživatelská příručka](docs/USER_GUIDE.md) | Poskytovatelé, kombinace, integrace CLI, nasazení
-[Referenční informace k API](docs/API_REFERENCE.md) | Všechny koncové body s příklady
-[MCP server](open-sse/mcp-server/README.md) | 16 nástrojů MCP, konfigurace IDE, klienti Python/TS/Go
-[Server A2A](src/lib/a2a/README.md) | Protokol JSON-RPC 2.0, dovednosti, streamování, správa úloh
-[Auto-Combo Engine](docs/auto-combo.md) | 6faktorové bodování, balíčky režimů, samoléčba
-[Odstraňování problémů](docs/TROUBLESHOOTING.md) | Běžné problémy a jejich řešení
-[Architektura](docs/ARCHITECTURE.md) | Architektura a interní prvky systému
-[Přispívání](CONTRIBUTING.md) | Nastavení a pokyny pro vývoj
-[Specifikace OpenAPI](docs/openapi.yaml) | Specifikace OpenAPI 3.0
-[Bezpečnostní zásady](SECURITY.md) | Hlášení zranitelností a bezpečnostní postupy
-[Nasazení virtuálního počítače](docs/VM_DEPLOYMENT_GUIDE.md) | Kompletní průvodce: Nastavení virtuálního počítače + nginx + Cloudflare
-[Galerie funkcí](docs/FEATURES.md) | Vizuální prohlídka řídicího panelu se snímky obrazovky
-[Kontrolní seznam vydání](docs/RELEASE_CHECKLIST.md) | Kroky ověření před vydáním
+| Dokument                                                     | Popis                                                                   |
+| ------------------------------------------------------------ | ----------------------------------------------------------------------- |
+| [Uživatelská příručka](docs/USER_GUIDE.md)                   | Poskytovatelé, kombinace, integrace CLI, nasazení                       |
+| [Referenční informace k API](docs/API_REFERENCE.md)          | Všechny koncové body s příklady                                         |
+| [MCP server](open-sse/mcp-server/README.md)                  | 16 nástrojů MCP, konfigurace IDE, klienti Python/TS/Go                  |
+| [Server A2A](src/lib/a2a/README.md)                          | Protokol JSON-RPC 2.0, dovednosti, streamování, správa úloh             |
+| [Auto-Combo Engine](docs/auto-combo.md)                      | 6faktorové bodování, balíčky režimů, samoléčba                          |
+| [Odstraňování problémů](docs/TROUBLESHOOTING.md)             | Běžné problémy a jejich řešení                                          |
+| [Architektura](docs/ARCHITECTURE.md)                         | Architektura a interní prvky systému                                    |
+| [Přispívání](CONTRIBUTING.md)                                | Nastavení a pokyny pro vývoj                                            |
+| [Specifikace OpenAPI](docs/openapi.yaml)                     | Specifikace OpenAPI 3.0                                                 |
+| [Bezpečnostní zásady](SECURITY.md)                           | Hlášení zranitelností a bezpečnostní postupy                            |
+| [Nasazení virtuálního počítače](docs/VM_DEPLOYMENT_GUIDE.md) | Kompletní průvodce: Nastavení virtuálního počítače + nginx + Cloudflare |
+| [Galerie funkcí](docs/FEATURES.md)                           | Vizuální prohlídka řídicího panelu se snímky obrazovky                  |
+| [Kontrolní seznam vydání](docs/RELEASE_CHECKLIST.md)         | Kroky ověření před vydáním                                              |
 
 ---
 
@@ -1938,14 +1814,14 @@ Dokument | Popis
 
 OmniRoute má **v plánu více než 210 funkcí** v několika fázích vývoje. Zde jsou klíčové oblasti:
 
-Kategorie | Plánované funkce | Hlavní body
---- | --- | ---
-🧠 **Směrování a inteligence** | 25+ | Směrování s nejnižší latencí, směrování založené na tagech, kontrola kvót před výstupem, výběr účtu P2C
-🔒 **Zabezpečení a dodržování předpisů** | 20+ | Zpevnění SSRF, maskování přihlašovacích údajů, limit rychlosti pro každý koncový bod, stanovení rozsahu klíčů pro správu
-📊 **Pozorovatelnost** | 15+ | Integrace OpenTelemetry, sledování kvót v reálném čase, sledování nákladů podle modelu
-🔄 **Integrace poskytovatelů** | 20+ | Dynamický registr modelů, doba zchlazení poskytovatelů, Codex pro více účtů, analýza kvót Copilota
-⚡ **Výkon** | 15+ | Dvojitá vrstva mezipaměti, mezipaměť výzev, mezipaměť odpovědí, udržování streamování, dávkové API
-🌐 **Ekosystém** | 10+ | WebSocket API, horké opětovné načítání konfigurace, distribuované úložiště konfigurace, komerční režim
+| Kategorie                                | Plánované funkce | Hlavní body                                                                                                              |
+| ---------------------------------------- | ---------------- | ------------------------------------------------------------------------------------------------------------------------ |
+| 🧠 **Směrování a inteligence**           | 25+              | Směrování s nejnižší latencí, směrování založené na tagech, kontrola kvót před výstupem, výběr účtu P2C                  |
+| 🔒 **Zabezpečení a dodržování předpisů** | 20+              | Zpevnění SSRF, maskování přihlašovacích údajů, limit rychlosti pro každý koncový bod, stanovení rozsahu klíčů pro správu |
+| 📊 **Pozorovatelnost**                   | 15+              | Integrace OpenTelemetry, sledování kvót v reálném čase, sledování nákladů podle modelu                                   |
+| 🔄 **Integrace poskytovatelů**           | 20+              | Dynamický registr modelů, doba zchlazení poskytovatelů, Codex pro více účtů, analýza kvót Copilota                       |
+| ⚡ **Výkon**                             | 15+              | Dvojitá vrstva mezipaměti, mezipaměť výzev, mezipaměť odpovědí, udržování streamování, dávkové API                       |
+| 🌐 **Ekosystém**                         | 10+              | WebSocket API, horké opětovné načítání konfigurace, distribuované úložiště konfigurace, komerční režim                   |
 
 ### 🔜 Již brzy
 

docs/i18n/da/README.md

@@ -934,8 +934,8 @@ OmniRoute v2.0 er bygget som en operationel platform, ikke kun en relæ-proxy.
 | Funktion | Hvad det gør || -------------------------- | -------------------------------------------------------------------- |
 | 🖼️ **Billedgenerering** | `/v1/images/generations` med cloud og lokale backends |
 | 📐 **Indlejringer** | `/v1/embeddings` til søgning og RAG-rørledninger |
-| 🎤 **Lydtransskription** | `/v1/audio/transcriptions` (Whisper og yderligere udbydere) |
-| 🔊 **Tekst-til-tale** | `/v1/audio/speech` (flere motorer/udbydere) |
+| 🎤 **Lydtransskription** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Tekst-til-tale** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise) |
 | 🎬 **Videogenerering** | `/v1/videos/generations` (ComfyUI + SD WebUI-arbejdsgange) |
 | 🎵 **Music Generation** | `/v1/music/generations` (ComfyUI-arbejdsgange) |
 | 🛡️ **Moderationer** | `/v1/moderations` sikkerhedstjek |

docs/i18n/de/README.md

@@ -939,8 +939,8 @@ OmniRoute v2.0 ist als Betriebsplattform konzipiert und nicht nur als Relay-Prox
 | Funktion | Was es tut || -------------------------- | ------------------------------------------------------------- |
 | 🖼️ **Bilderzeugung** | `/v1/images/generations` mit Cloud- und lokalen Backends |
 | 📐 **Einbettungen** | `/v1/embeddings` für Such- und RAG-Pipelines |
-| 🎤 **Audio-Transkription** | `/v1/audio/transcriptions` (Whisper und zusätzliche Anbieter) |
-| 🔊 **Text-to-Speech** | `/v1/audio/speech` (mehrere Engines/Anbieter) |
+| 🎤 **Audio-Transkription** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Text-to-Speech** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise) |
 | 🎬 **Videogenerierung** | `/v1/videos/generations` (ComfyUI + SD WebUI-Workflows) |
 | 🎵 **Musikgeneration** | `/v1/music/generations` (ComfyUI-Workflows) |
 | 🛡️ **Moderationen** | `/v1/moderations` Sicherheitsprüfungen |

docs/i18n/es/README.md

@@ -877,14 +877,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 APIs Multi-Modal
 
-| Característica                | Qué Hace                                               |
-| ----------------------------- | ------------------------------------------------------ |
-| 🖼️ **Generación de Imágenes** | `/v1/images/generations` — 4 proveedores, 9+ modelos   |
-| 📐 **Embeddings**             | `/v1/embeddings` — 6 proveedores, 9+ modelos           |
-| 🎤 **Transcripción de Audio** | `/v1/audio/transcriptions` — Compatible con Whisper    |
-| 🔊 **Texto a Voz**            | `/v1/audio/speech` — Síntesis de audio multi-proveedor |
-| 🛡️ **Moderaciones**           | `/v1/moderations` — Verificaciones de seguridad        |
-| 🔀 **Reranking**              | `/v1/rerank` — Reranking de relevancia de documentos   |
+| Característica                | Qué Hace                                                                                                                                                                   |
+| ----------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Generación de Imágenes** | `/v1/images/generations` — 4 proveedores, 9+ modelos                                                                                                                       |
+| 📐 **Embeddings**             | `/v1/embeddings` — 6 proveedores, 9+ modelos                                                                                                                               |
+| 🎤 **Transcripción de Audio** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Texto a Voz**            | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderaciones**           | `/v1/moderations` — Verificaciones de seguridad                                                                                                                            |
+| 🔀 **Reranking**              | `/v1/rerank` — Reranking de relevancia de documentos                                                                                                                       |
 
 ### 🛡️ Resiliencia y Seguridad
 

docs/i18n/fi/README.md

@@ -874,14 +874,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Multimodaaliset sovellusliittymät
 
-| Ominaisuus                | Mitä se tekee                                             |
-| ------------------------- | --------------------------------------------------------- |
-| 🖼️ **Kuvan luominen**     | `/v1/images/generations` — 4 toimittajaa, 9+ mallia       |
-| 📐 **Upotukset**          | `/v1/embeddings` — 6 toimittajaa, 9+ mallia               |
-| 🎤 **Äänitranskriptio**   | `/v1/audio/transcriptions` — Kuiskausyhteensopiva         |
-| 🔊 **Tekstistä puheeksi** | `/v1/audio/speech` — Usean palveluntarjoajan äänisynteesi |
-| 🛡️ **Moderaatiot**        | `/v1/moderations` — Sisällön turvallisuustarkistukset     |
-| 🔀 **Uudelleenjärjestys** | `/v1/rerank` — Asiakirjan osuvuuden uudelleensijoitus     |
+| Ominaisuus                | Mitä se tekee                                                                                                                                                              |
+| ------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Kuvan luominen**     | `/v1/images/generations` — 4 toimittajaa, 9+ mallia                                                                                                                        |
+| 📐 **Upotukset**          | `/v1/embeddings` — 6 toimittajaa, 9+ mallia                                                                                                                                |
+| 🎤 **Äänitranskriptio**   | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Tekstistä puheeksi** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderaatiot**        | `/v1/moderations` — Sisällön turvallisuustarkistukset                                                                                                                      |
+| 🔀 **Uudelleenjärjestys** | `/v1/rerank` — Asiakirjan osuvuuden uudelleensijoitus                                                                                                                      |
 
 ### 🛡️ Joustavuus ja turvallisuus
 

docs/i18n/fr/README.md

@@ -875,14 +875,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 APIs multi-modales
 
-| Fonctionnalité             | Ce qu'elle fait                                         |
-| -------------------------- | ------------------------------------------------------- |
-| 🖼️ **Génération d'images** | `/v1/images/generations` — 4 fournisseurs, 9+ modèles   |
-| 📐 **Embeddings**          | `/v1/embeddings` — 6 fournisseurs, 9+ modèles           |
-| 🎤 **Transcription audio** | `/v1/audio/transcriptions` — compatible Whisper         |
-| 🔊 **Texte vers parole**   | `/v1/audio/speech` — synthèse audio multi-fournisseur   |
-| 🛡️ **Modérations**         | `/v1/moderations` — vérifications de sécurité           |
-| 🔀 **Reranking**           | `/v1/rerank` — reclassement de pertinence des documents |
+| Fonctionnalité             | Ce qu'elle fait                                                                                                                                                            |
+| -------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Génération d'images** | `/v1/images/generations` — 4 fournisseurs, 9+ modèles                                                                                                                      |
+| 📐 **Embeddings**          | `/v1/embeddings` — 6 fournisseurs, 9+ modèles                                                                                                                              |
+| 🎤 **Transcription audio** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Texte vers parole**   | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Modérations**         | `/v1/moderations` — vérifications de sécurité                                                                                                                              |
+| 🔀 **Reranking**           | `/v1/rerank` — reclassement de pertinence des documents                                                                                                                    |
 
 ### 🛡️ Résilience & Sécurité
 

docs/i18n/he/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 ממשקי API רב-מודאליים
 
-| תכונה               | מה זה עושה                                    |
-| ------------------- | --------------------------------------------- |
-| 🖼️ **יצירת תמונות** | `/v1/images/generations` — 4 ספקים, 9+ דגמים  |
-| 📐 **הטבעות**       | `/v1/embeddings` — 6 ספקים, 9+ דגמים          |
-| 🎤 **תמלול אודיו**  | `/v1/audio/transcriptions` — תואם לחישה       |
-| 🔊 **טקסט לדיבור**  | `/v1/audio/speech` — סינתזת אודיו מרובה ספקים |
-| 🛡️ **מנחים**        | `/v1/moderations` — בדיקות בטיחות תוכן        |
-| 🔀 **דירוג מחדש**   | `/v1/rerank` — דירוג מחדש של רלוונטיות המסמך  |
+| תכונה               | מה זה עושה                                                                                                                                                                 |
+| ------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **יצירת תמונות** | `/v1/images/generations` — 4 ספקים, 9+ דגמים                                                                                                                               |
+| 📐 **הטבעות**       | `/v1/embeddings` — 6 ספקים, 9+ דגמים                                                                                                                                       |
+| 🎤 **תמלול אודיו**  | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **טקסט לדיבור**  | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **מנחים**        | `/v1/moderations` — בדיקות בטיחות תוכן                                                                                                                                     |
+| 🔀 **דירוג מחדש**   | `/v1/rerank` — דירוג מחדש של רלוונטיות המסמך                                                                                                                               |
 
 ### 🛡️ חוסן וביטחון
 

docs/i18n/hu/README.md

@@ -874,14 +874,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Multimodális API-k
 
-| Funkció                | Mit csinál                                                   |
-| ---------------------- | ------------------------------------------------------------ |
-| 🖼️ **Képgenerálás**    | `/v1/images/generations` — 4 szolgáltató, 9+ modell          |
-| 📐 **Beágyazás**       | `/v1/embeddings` — 6 szolgáltató, 9+ modell                  |
-| 🎤 **Audio átírás**    | `/v1/audio/transcriptions` — Suttogás-kompatibilis           |
-| 🔊 **Szövegfelolvasó** | `/v1/audio/speech` — Hangszintézis több szolgáltatónál       |
-| 🛡️ **Moderálás**       | `/v1/moderations` — Tartalombiztonsági ellenőrzések          |
-| 🔀 **Átsorolás**       | `/v1/rerank` — A dokumentumok relevancia szerinti átsorolása |
+| Funkció                | Mit csinál                                                                                                                                                                 |
+| ---------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Képgenerálás**    | `/v1/images/generations` — 4 szolgáltató, 9+ modell                                                                                                                        |
+| 📐 **Beágyazás**       | `/v1/embeddings` — 6 szolgáltató, 9+ modell                                                                                                                                |
+| 🎤 **Audio átírás**    | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Szövegfelolvasó** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderálás**       | `/v1/moderations` — Tartalombiztonsági ellenőrzések                                                                                                                        |
+| 🔀 **Átsorolás**       | `/v1/rerank` — A dokumentumok relevancia szerinti átsorolása                                                                                                               |
 
 ### 🛡️ Rugalmasság és biztonság
 

docs/i18n/id/README.md

@@ -874,14 +874,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 API Multi-Modal
 
-| Fitur                      | Apa Fungsinya                                          |
-| -------------------------- | ------------------------------------------------------ |
-| 🖼️ **Pembuatan Gambar**    | `/v1/images/generations` — 4 penyedia, 9+ model        |
-| 📐 **Sematan**             | `/v1/embeddings` — 6 penyedia, 9+ model                |
-| 🎤 **Transkripsi Audio**   | `/v1/audio/transcriptions` — Kompatibel dengan bisikan |
-| 🔊 **Teks-ke-Ucapan**      | `/v1/audio/speech` — Sintesis audio multi-penyedia     |
-| 🛡️ **Moderasi**            | `/v1/moderations` — Pemeriksaan keamanan konten        |
-| 🔀 **Pemeringkatan Ulang** | `/v1/rerank` — Pemeringkatan ulang relevansi dokumen   |
+| Fitur                      | Apa Fungsinya                                                                                                                                                              |
+| -------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Pembuatan Gambar**    | `/v1/images/generations` — 4 penyedia, 9+ model                                                                                                                            |
+| 📐 **Sematan**             | `/v1/embeddings` — 6 penyedia, 9+ model                                                                                                                                    |
+| 🎤 **Transkripsi Audio**   | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Teks-ke-Ucapan**      | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderasi**            | `/v1/moderations` — Pemeriksaan keamanan konten                                                                                                                            |
+| 🔀 **Pemeringkatan Ulang** | `/v1/rerank` — Pemeringkatan ulang relevansi dokumen                                                                                                                       |
 
 ### 🛡️ Ketahanan & Keamanan
 

docs/i18n/in/README.md

@@ -770,14 +770,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 मल्टी-मॉडल एपीआई
 
-| फ़ीचर                        | यह क्या करता है                                   |
-| ---------------------------- | ------------------------------------------------- |
-| 🖼️ **छवि निर्माण**           | `/v1/images/generations` - 4 प्रदाता, 9+ मॉडल     |
-| 📐 **एंबेडिंग**              | `/v1/embeddings` — 6 प्रदाता, 9+ मॉडल             |
-| 🎤 **ऑडियो ट्रांस्क्रिप्शन** | `/v1/audio/transcriptions` - कानाफूसी-संगत        |
-| 🔊 **टेक्स्ट-टू-स्पीच**      | `/v1/audio/speech` - बहु-प्रदाता ऑडियो संश्लेषण   |
-| 🛡️ **संयम**                  | `/v1/moderations` — सामग्री सुरक्षा जांच          |
-| 🔀 **पुनर्रैंकिंग**          | `/v1/rerank` — दस्तावेज़ प्रासंगिकता पुनर्रैंकिंग |
+| फ़ीचर                        | यह क्या करता है                                                                                                                                                            |
+| ---------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **छवि निर्माण**           | `/v1/images/generations` - 4 प्रदाता, 9+ मॉडल                                                                                                                              |
+| 📐 **एंबेडिंग**              | `/v1/embeddings` — 6 प्रदाता, 9+ मॉडल                                                                                                                                      |
+| 🎤 **ऑडियो ट्रांस्क्रिप्शन** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **टेक्स्ट-टू-स्पीच**      | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **संयम**                  | `/v1/moderations` — सामग्री सुरक्षा जांच                                                                                                                                   |
+| 🔀 **पुनर्रैंकिंग**          | `/v1/rerank` — दस्तावेज़ प्रासंगिकता पुनर्रैंकिंग                                                                                                                          |
 
 ### 🛡️ लचीलापन और सुरक्षा
 

docs/i18n/it/README.md

@@ -874,14 +874,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 API Multi-modali
 
-| Funzionalità                | Cosa Fa                                              |
-| --------------------------- | ---------------------------------------------------- |
-| 🖼️ **Generazione immagini** | `/v1/images/generations` — 4 provider, 9+ modelli    |
-| 📐 **Embeddings**           | `/v1/embeddings` — 6 provider, 9+ modelli            |
-| 🎤 **Trascrizione audio**   | `/v1/audio/transcriptions` — Compatibile Whisper     |
-| 🔊 **Testo a voce**         | `/v1/audio/speech` — Sintesi audio multi-provider    |
-| 🛡️ **Moderazioni**          | `/v1/moderations` — Controlli di sicurezza           |
-| 🔀 **Reranking**            | `/v1/rerank` — Riclassificazione rilevanza documenti |
+| Funzionalità                | Cosa Fa                                                                                                                                                                    |
+| --------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Generazione immagini** | `/v1/images/generations` — 4 provider, 9+ modelli                                                                                                                          |
+| 📐 **Embeddings**           | `/v1/embeddings` — 6 provider, 9+ modelli                                                                                                                                  |
+| 🎤 **Trascrizione audio**   | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Testo a voce**         | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderazioni**          | `/v1/moderations` — Controlli di sicurezza                                                                                                                                 |
+| 🔀 **Reranking**            | `/v1/rerank` — Riclassificazione rilevanza documenti                                                                                                                       |
 
 ### 🛡️ Resilienza & Sicurezza
 

docs/i18n/ja/README.md

@@ -874,14 +874,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 マルチモーダル API
 
-| 特集                    | 何をするのか                                                    |
-| ----------------------- | --------------------------------------------------------------- |
-| 🖼️ **画像生成**         | `/v1/images/generations` — 4 つのプロバイダー、9 つ以上のモデル |
-| 📐 **埋め込み**         | `/v1/embeddings` — 6 つのプロバイダー、9 つ以上のモデル         |
-| 🎤 **音声文字起こし**   | `/v1/audio/transcriptions` — ウィスパー互換                     |
-| 🔊 **テキスト読み上げ** | `/v1/audio/speech` — マルチプロバイダーのオーディオ合成         |
-| 🛡️ **モデレーション**   | `/v1/moderations` — コンテンツの安全性チェック                  |
-| 🔀 **再ランキング**     | `/v1/rerank` — ドキュメントの関連性の再ランキング               |
+| 特集                    | 何をするのか                                                                                                                                                               |
+| ----------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **画像生成**         | `/v1/images/generations` — 4 つのプロバイダー、9 つ以上のモデル                                                                                                            |
+| 📐 **埋め込み**         | `/v1/embeddings` — 6 つのプロバイダー、9 つ以上のモデル                                                                                                                    |
+| 🎤 **音声文字起こし**   | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **テキスト読み上げ** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **モデレーション**   | `/v1/moderations` — コンテンツの安全性チェック                                                                                                                             |
+| 🔀 **再ランキング**     | `/v1/rerank` — ドキュメントの関連性の再ランキング                                                                                                                          |
 
 ### 🛡️ 復元力とセキュリティ
 

docs/i18n/ko/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 다중 모드 API
 
-| 기능                    | 그것이 하는 일                                         |
-| ----------------------- | ------------------------------------------------------ |
-| 🖼️ **이미지 생성**      | `/v1/images/generations` — 4개 공급자, 9개 이상의 모델 |
-| 📐 **임베딩**           | `/v1/embeddings` — 6개 공급자, 9개 이상의 모델         |
-| 🎤 **오디오 전사**      | `/v1/audio/transcriptions` — 속삭임 호환               |
-| 🔊 **텍스트 음성 변환** | `/v1/audio/speech` — 다중 제공자 오디오 합성           |
-| 🛡️ **조정**             | `/v1/moderations` — 콘텐츠 안전 확인                   |
-| 🔀 **재순위**           | `/v1/rerank` — 문서 관련성 재순위                      |
+| 기능                    | 그것이 하는 일                                                                                                                                                             |
+| ----------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **이미지 생성**      | `/v1/images/generations` — 4개 공급자, 9개 이상의 모델                                                                                                                     |
+| 📐 **임베딩**           | `/v1/embeddings` — 6개 공급자, 9개 이상의 모델                                                                                                                             |
+| 🎤 **오디오 전사**      | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **텍스트 음성 변환** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **조정**             | `/v1/moderations` — 콘텐츠 안전 확인                                                                                                                                       |
+| 🔀 **재순위**           | `/v1/rerank` — 문서 관련성 재순위                                                                                                                                          |
 
 ### 🛡️ 복원력 및 보안
 

docs/i18n/ms/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 API Berbilang Modal
 
-| Ciri                     | Apa yang Dilakukan                                     |
-| ------------------------ | ------------------------------------------------------ |
-| 🖼️ **Penjanaan Imej**    | `/v1/images/generations` — 4 pembekal, 9+ model        |
-| 📐 **Pembenaman**        | `/v1/embeddings` — 6 pembekal, 9+ model                |
-| 🎤 **Transkripsi Audio** | `/v1/audio/transcriptions` — Serasi dengan bisikan     |
-| 🔊 **Teks-ke-Ucapan**    | `/v1/audio/speech` — Sintesis audio berbilang pembekal |
-| 🛡️ **Kesederhanaan**     | `/v1/moderations` — Pemeriksaan keselamatan kandungan  |
-| 🔀 **Penyusunan semula** | `/v1/rerank` — Penarafan semula perkaitan dokumen      |
+| Ciri                     | Apa yang Dilakukan                                                                                                                                                         |
+| ------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Penjanaan Imej**    | `/v1/images/generations` — 4 pembekal, 9+ model                                                                                                                            |
+| 📐 **Pembenaman**        | `/v1/embeddings` — 6 pembekal, 9+ model                                                                                                                                    |
+| 🎤 **Transkripsi Audio** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Teks-ke-Ucapan**    | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Kesederhanaan**     | `/v1/moderations` — Pemeriksaan keselamatan kandungan                                                                                                                      |
+| 🔀 **Penyusunan semula** | `/v1/rerank` — Penarafan semula perkaitan dokumen                                                                                                                          |
 
 ### 🛡️ Ketahanan & Keselamatan
 

docs/i18n/nl/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Multimodale API's
 
-| Kenmerk                  | Wat het doet                                              |
-| ------------------------ | --------------------------------------------------------- |
-| 🖼️ **Beeldgeneratie**    | `/v1/images/generations` — 4 providers, 9+ modellen       |
-| 📐 **Insluitingen**      | `/v1/embeddings` — 6 providers, 9+ modellen               |
-| 🎤 **Audiotranscriptie** | `/v1/audio/transcriptions` — Whisper-compatibel           |
-| 🔊 **Tekst-naar-spraak** | `/v1/audio/speech` — Audiosynthese van meerdere providers |
-| 🛡️ **Moderaties**        | `/v1/moderations` — Veiligheidscontroles van inhoud       |
-| 🔀 **Herschikking**      | `/v1/rerank` — Herschikking van documentrelevantie        |
+| Kenmerk                  | Wat het doet                                                                                                                                                               |
+| ------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Beeldgeneratie**    | `/v1/images/generations` — 4 providers, 9+ modellen                                                                                                                        |
+| 📐 **Insluitingen**      | `/v1/embeddings` — 6 providers, 9+ modellen                                                                                                                                |
+| 🎤 **Audiotranscriptie** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Tekst-naar-spraak** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderaties**        | `/v1/moderations` — Veiligheidscontroles van inhoud                                                                                                                        |
+| 🔀 **Herschikking**      | `/v1/rerank` — Herschikking van documentrelevantie                                                                                                                         |
 
 ### 🛡️ Veerkracht en veiligheid
 

docs/i18n/no/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Multi-Modal APIer
 
-| Funksjon                | Hva det gjør                                           |
-| ----------------------- | ------------------------------------------------------ |
-| 🖼️ **Bildegenerering**  | `/v1/images/generations` — 4 leverandører, 9+ modeller |
-| 📐 **Innbygging**       | `/v1/embeddings` — 6 leverandører, 9+ modeller         |
-| 🎤 **Lydtranskripsjon** | `/v1/audio/transcriptions` — Whisper-kompatibel        |
-| 🔊 **Tekst-til-tale**   | `/v1/audio/speech` — Multi-leverandør lydsyntese       |
-| 🛡️ **Moderasjoner**     | `/v1/moderations` — Innholdssikkerhetssjekker          |
-| 🔀 **Omrangering**      | `/v1/rerank` — Rerangering av dokumentrelevans         |
+| Funksjon                | Hva det gjør                                                                                                                                                               |
+| ----------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Bildegenerering**  | `/v1/images/generations` — 4 leverandører, 9+ modeller                                                                                                                     |
+| 📐 **Innbygging**       | `/v1/embeddings` — 6 leverandører, 9+ modeller                                                                                                                             |
+| 🎤 **Lydtranskripsjon** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Tekst-til-tale**   | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderasjoner**     | `/v1/moderations` — Innholdssikkerhetssjekker                                                                                                                              |
+| 🔀 **Omrangering**      | `/v1/rerank` — Rerangering av dokumentrelevans                                                                                                                             |
 
 ### 🛡️ Spenst og sikkerhet
 

docs/i18n/phi/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Mga Multi-Modal na API
 
-| Tampok                     | Ano ang Ginagawa Nito                                        |
-| -------------------------- | ------------------------------------------------------------ |
-| 🖼️ **Pagbuo ng Larawan**   | `/v1/images/generations` — 4 na provider, 9+ na modelo       |
-| 📐 **Mga Pag-embed**       | `/v1/embeddings` — 6 na provider, 9+ na modelo               |
-| 🎤 **Audio Transcription** | `/v1/audio/transcriptions` — Whisper-compatible              |
-| 🔊 **Text-to-Speech**      | `/v1/audio/speech` — Multi-provider audio synthesis          |
-| 🛡️ **Mga Pag-moderate**    | `/v1/moderations` — Mga pagsusuri sa kaligtasan ng nilalaman |
-| 🔀 **Reranking**           | `/v1/rerank` — Muling pagraranggo ng kaugnayan ng dokumento  |
+| Tampok                     | Ano ang Ginagawa Nito                                                                                                                                                      |
+| -------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Pagbuo ng Larawan**   | `/v1/images/generations` — 4 na provider, 9+ na modelo                                                                                                                     |
+| 📐 **Mga Pag-embed**       | `/v1/embeddings` — 6 na provider, 9+ na modelo                                                                                                                             |
+| 🎤 **Audio Transcription** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Text-to-Speech**      | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Mga Pag-moderate**    | `/v1/moderations` — Mga pagsusuri sa kaligtasan ng nilalaman                                                                                                               |
+| 🔀 **Reranking**           | `/v1/rerank` — Muling pagraranggo ng kaugnayan ng dokumento                                                                                                                |
 
 ### 🛡️ Katatagan at Seguridad
 

docs/i18n/pl/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Wielomodalne interfejsy API
 
-| Funkcja                       | Co to robi                                             |
-| ----------------------------- | ------------------------------------------------------ |
-| 🖼️ **Generowanie obrazu**     | `/v1/images/generations` — 4 dostawców, ponad 9 modeli |
-| 📐 **Osadzenia**              | `/v1/embeddings` — 6 dostawców, ponad 9 modeli         |
-| 🎤 **Transkrypcja audio**     | `/v1/audio/transcriptions` — Kompatybilny z szeptem    |
-| 🔊 **Zamiana tekstu na mowę** | `/v1/audio/speech` — Synteza dźwięku wielu dostawców   |
-| 🛡️ **Moderacje**              | `/v1/moderations` — Kontrola bezpieczeństwa treści     |
-| 🔀 **Ponowna pozycja**        | `/v1/rerank` — Zmiana rankingu trafności dokumentu     |
+| Funkcja                       | Co to robi                                                                                                                                                                 |
+| ----------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Generowanie obrazu**     | `/v1/images/generations` — 4 dostawców, ponad 9 modeli                                                                                                                     |
+| 📐 **Osadzenia**              | `/v1/embeddings` — 6 dostawców, ponad 9 modeli                                                                                                                             |
+| 🎤 **Transkrypcja audio**     | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Zamiana tekstu na mowę** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderacje**              | `/v1/moderations` — Kontrola bezpieczeństwa treści                                                                                                                         |
+| 🔀 **Ponowna pozycja**        | `/v1/rerank` — Zmiana rankingu trafności dokumentu                                                                                                                         |
 
 ### 🛡️ Odporność i bezpieczeństwo
 

docs/i18n/pt-BR/README.md

@@ -819,24 +819,28 @@ Quando minimizado, o OmniRoute fica na bandeja do sistema com ações rápidas:
 
 ## 💰 Preços Resumidos
 
-| Tier              | Provedor          | Custo                        | Reset de Cota     | Melhor Para             |
-| ----------------- | ----------------- | ---------------------------- | ----------------- | ----------------------- |
-| **💳 ASSINATURA** | Claude Code (Pro) | $20/mês                      | 5h + semanal      | Já é assinante          |
-|                   | Codex (Plus/Pro)  | $20-200/mês                  | 5h + semanal      | Usuários OpenAI         |
-|                   | Gemini CLI        | **GRATUITO**                 | 180K/mês + 1K/dia | Todos!                  |
-|                   | GitHub Copilot    | $10-19/mês                   | Mensal            | Usuários GitHub         |
-| **🔑 API KEY**    | NVIDIA NIM        | **GRATUITO** (1000 créditos) | Único             | Testes gratuitos        |
-|                   | DeepSeek          | Por uso                      | Nenhum            | Melhor preço/qualidade  |
-|                   | Groq              | Tier gratuito + pago         | Limitado          | Inferência ultra-rápida |
-|                   | xAI (Grok)        | Por uso                      | Nenhum            | Modelos Grok            |
-|                   | Mistral           | Tier gratuito + pago         | Limitado          | IA Europeia             |
-|                   | OpenRouter        | Por uso                      | Nenhum            | 100+ modelos            |
-| **💰 BARATO**     | GLM-4.7           | $0.6/1M                      | Diário 10h        | Backup econômico        |
-|                   | MiniMax M2.1      | $0.2/1M                      | Rotativo 5h       | Opção mais barata       |
-|                   | Kimi K2           | $9/mês fixo                  | 10M tokens/mês    | Custo previsível        |
-| **🆓 GRATUITO**   | iFlow             | $0                           | Ilimitado         | 8 modelos gratuitos     |
-|                   | Qwen              | $0                           | Ilimitado         | 3 modelos gratuitos     |
-|                   | Kiro              | $0                           | Ilimitado         | Claude gratuito         |
+| Tier              | Provedor          | Custo                        | Reset de Cota     | Melhor Para                    |
+| ----------------- | ----------------- | ---------------------------- | ----------------- | ------------------------------ |
+| **💳 ASSINATURA** | Claude Code (Pro) | $20/mês                      | 5h + semanal      | Já é assinante                 |
+|                   | Codex (Plus/Pro)  | $20-200/mês                  | 5h + semanal      | Usuários OpenAI                |
+|                   | Gemini CLI        | **GRATUITO**                 | 180K/mês + 1K/dia | Todos!                         |
+|                   | GitHub Copilot    | $10-19/mês                   | Mensal            | Usuários GitHub                |
+| **🔑 API KEY**    | NVIDIA NIM        | **GRATUITO** (1000 créditos) | Único             | Testes gratuitos               |
+|                   | DeepSeek          | Por uso                      | Nenhum            | Melhor preço/qualidade         |
+|                   | Groq              | Tier gratuito + pago         | Limitado          | Inferência ultra-rápida        |
+|                   | xAI (Grok)        | Por uso                      | Nenhum            | Modelos Grok                   |
+|                   | Mistral           | Tier gratuito + pago         | Limitado          | IA Europeia                    |
+|                   | OpenRouter        | Por uso                      | Nenhum            | 100+ modelos                   |
+| **💰 BARATO**     | GLM-4.7           | $0.6/1M                      | Diário 10h        | Backup econômico               |
+|                   | MiniMax M2.1      | $0.2/1M                      | Rotativo 5h       | Opção mais barata              |
+|                   | Kimi K2           | $9/mês fixo                  | 10M tokens/mês    | Custo previsível               |
+| **🆓 GRATUITO**   | iFlow             | $0                           | Ilimitado         | 8 modelos gratuitos            |
+|                   | Qwen              | $0                           | Ilimitado         | 3 modelos gratuitos            |
+|                   | Kiro              | $0                           | Ilimitado         | Claude gratuito                |
+|                   | LongCat 🆕        | **$0** (50M tok/dia 🔥)      | 1 req/s           | Maior cota grátis do mundo     |
+|                   | Pollinations 🆕   | **$0** (sem chave API)       | 1 req/15s         | GPT-5, Claude, DeepSeek, Llama |
+|                   | Cloudflare AI 🆕  | **$0** (10K Neurons/dia)     | ~150 resp/dia     | 50+ modelos, edge global       |
+|                   | Scaleway AI 🆕    | **$0** (1M tokens total)     | Limitado por taxa | EU/GDPR, Qwen3 235B, Llama 70B |
 
 **💡 Dica Pro:** Comece com Gemini CLI (180K grátis/mês) + iFlow (ilimitado grátis) = $0 de custo!
 
@@ -879,16 +883,16 @@ Por que isso é relevante:
 
 ### 🎵 APIs Multi-Modal
 
-| Funcionalidade              | O que Faz                                                                                                   |
-| --------------------------- | ----------------------------------------------------------------------------------------------------------- |
-| 🖼️ **Geração de Imagem**    | `/v1/images/generations` — 10 provedores, 20+ modelos (cloud + local)                                       |
-| 📐 **Embeddings**           | `/v1/embeddings` — 6 provedores, 9+ modelos                                                                 |
-| 🎤 **Transcrição de Áudio** | `/v1/audio/transcriptions` — Whisper + Nvidia NIM, HuggingFace, Qwen3                                       |
-| 🔊 **Texto para Fala**      | `/v1/audio/speech` — ElevenLabs, Nvidia NIM, HuggingFace, Coqui, Tortoise, Qwen3, Inworld, Cartesia, PlayHT |
-| 🎬 **Geração de Vídeo**     | `/v1/videos/generations` — ComfyUI (AnimateDiff, SVD), SD WebUI                                             |
-| 🎵 **Geração de Música**    | `/v1/music/generations` — ComfyUI (Stable Audio Open, MusicGen)                                             |
-| 🛡️ **Moderações**           | `/v1/moderations` — Verificações de segurança                                                               |
-| 🔀 **Reranking**            | `/v1/rerank` — Reranking de relevância de documentos                                                        |
+| Funcionalidade              | O que Faz                                                                                                                                                                  |
+| --------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Geração de Imagem**    | `/v1/images/generations` — 10 provedores, 20+ modelos (cloud + local)                                                                                                      |
+| 📐 **Embeddings**           | `/v1/embeddings` — 6 provedores, 9+ modelos                                                                                                                                |
+| 🎤 **Transcrição de Áudio** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Texto para Fala**      | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🎬 **Geração de Vídeo**     | `/v1/videos/generations` — ComfyUI (AnimateDiff, SVD), SD WebUI                                                                                                            |
+| 🎵 **Geração de Música**    | `/v1/music/generations` — ComfyUI (Stable Audio Open, MusicGen)                                                                                                            |
+| 🛡️ **Moderações**           | `/v1/moderations` — Verificações de segurança                                                                                                                              |
+| 🔀 **Reranking**            | `/v1/rerank` — Reranking de relevância de documentos                                                                                                                       |
 
 ### 🛡️ Resiliência e Segurança
 
@@ -1223,6 +1227,53 @@ Modelos:
   kr/claude-haiku-4.5
 ```
 
+### LongCat AI (GRATUITO 50M tokens/dia!) 🆕
+
+1. Cadastre-se: [longcat.chat](https://longcat.chat) com e-mail ou telefone
+2. Gere uma chave de API gratuita
+3. Dashboard → Adicionar Provedor → LongCat
+
+**Modelos:**
+
+- `lc/LongCat-Flash-Lite` — **50M tokens/dia** 💥 (maior cota gratuita do mundo!)
+- `lc/LongCat-Flash-Chat` — 500K tokens/dia
+- `lc/LongCat-Flash-Thinking` — 500K tokens/dia (raciocínio)
+
+> 100% gratuito durante o beta público. Reset diário à meia-noite UTC.
+
+### Pollinations AI (SEM CHAVE NECESSÁRIA!) 🆕
+
+1. Adicione o provedor Pollinations no Dashboard
+2. Deixe o campo de chave API vazio (ou coloque qualquer string)
+3. Comece a usar imediatamente!
+
+**Modelos via `pol/`:** `openai` (GPT-5), `claude`, `gemini`, `deepseek`, `llama` (Llama 4)
+
+> Sem cadastro, sem chave, sem cartão de crédito. 1 req/15s ilimitado.
+
+### Cloudflare Workers AI (GRATUITO 10K Neurons/dia!) 🆕
+
+1. Cadastre-se: [dash.cloudflare.com](https://dash.cloudflare.com)
+2. Gere um API Token em Profile → API Tokens
+3. Copie seu Account ID (coluna direita do dashboard)
+4. Dashboard → Adicionar Provedor → Cloudflare AI
+   - API Key: seu token
+   - Account ID: seu account ID
+
+**Modelos via `cf/`:** `@cf/meta/llama-3.3-70b-instruct`, `@cf/google/gemma-3-12b-it`, 50+ mais
+
+> 10K Neurons/dia ≈ 150 respostas de LLM ou 500s de transcrição Whisper gratuita!
+
+### Scaleway AI (1M tokens gratuitos!) 🆕
+
+1. Cadastre-se: [console.scaleway.com](https://console.scaleway.com)
+2. Gere uma chave de API IAM
+3. Dashboard → Adicionar Provedor → Scaleway
+
+**Modelos via `scw/`:** `qwen3-235b-a22b-instruct-2507` (Qwen3 235B!), `llama-3.1-70b-instruct`
+
+> 1M tokens gratuitos para novas contas. Dados processados na 🇫🇷 França (EU/GDPR).
+
 </details>
 
 <details>

docs/i18n/pt/README.md

@@ -874,14 +874,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 APIs multimodais
 
-| Recurso                           | O que faz                                                   |
-| --------------------------------- | ----------------------------------------------------------- |
-| 🖼️ **Geração de imagens**         | `/v1/images/generations` — 4 provedores, mais de 9 modelos  |
-| 📐 **Incorporações**              | `/v1/embeddings` — 6 provedores, mais de 9 modelos          |
-| 🎤 **Transcrição de áudio**       | `/v1/audio/transcriptions` — Compatível com sussurro        |
-| 🔊 **Conversão de texto em fala** | `/v1/audio/speech` — Síntese de áudio multiprovedor         |
-| 🛡️ **Moderações**                 | `/v1/moderations` — Verificações de segurança de conteúdo   |
-| 🔀 **Reclassificação**            | `/v1/rerank` — Reclassificação da relevância dos documentos |
+| Recurso                           | O que faz                                                                                                                                                                  |
+| --------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Geração de imagens**         | `/v1/images/generations` — 4 provedores, mais de 9 modelos                                                                                                                 |
+| 📐 **Incorporações**              | `/v1/embeddings` — 6 provedores, mais de 9 modelos                                                                                                                         |
+| 🎤 **Transcrição de áudio**       | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Conversão de texto em fala** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderações**                 | `/v1/moderations` — Verificações de segurança de conteúdo                                                                                                                  |
+| 🔀 **Reclassificação**            | `/v1/rerank` — Reclassificação da relevância dos documentos                                                                                                                |
 
 ### 🛡️ Resiliência e segurança
 

docs/i18n/ro/README.md

@@ -875,14 +875,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 API-uri multimodale
 
-| Caracteristica            | Ce face                                                    |
-| ------------------------- | ---------------------------------------------------------- |
-| 🖼️ **Generarea imaginii** | `/v1/images/generations` — 4 furnizori, peste 9 modele     |
-| 📐 **Inglobări**          | `/v1/embeddings` — 6 furnizori, peste 9 modele             |
-| 🎤 **Transcriere audio**  | `/v1/audio/transcriptions` — Compatibil cu Whisper         |
-| 🔊 **Text-to-speech**     | `/v1/audio/speech` — Sinteză audio cu mai mulți furnizori  |
-| 🛡️ **Moderații**          | `/v1/moderations` — Verificări de siguranță a conținutului |
-| 🔀 **Reclasificare**      | `/v1/rerank` — Reclasificarea relevanței documentului      |
+| Caracteristica            | Ce face                                                                                                                                                                    |
+| ------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Generarea imaginii** | `/v1/images/generations` — 4 furnizori, peste 9 modele                                                                                                                     |
+| 📐 **Inglobări**          | `/v1/embeddings` — 6 furnizori, peste 9 modele                                                                                                                             |
+| 🎤 **Transcriere audio**  | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Text-to-speech**     | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderații**          | `/v1/moderations` — Verificări de siguranță a conținutului                                                                                                                 |
+| 🔀 **Reclasificare**      | `/v1/rerank` — Reclasificarea relevanței documentului                                                                                                                      |
 
 ### 🛡️ Reziliență și securitate
 

docs/i18n/ru/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Мультимодальные API
 
-| Функция                      | Что делает                                          |
-| ---------------------------- | --------------------------------------------------- |
-| 🖼️ **Генерация изображений** | `/v1/images/generations` — 4 провайдера, 9+ моделей |
-| 📐 **Embeddings**            | `/v1/embeddings` — 6 провайдеров, 9+ моделей        |
-| 🎤 **Транскрипция аудио**    | `/v1/audio/transcriptions` — Совместимо с Whisper   |
-| 🔊 **Текст в речь**          | `/v1/audio/speech` — Мульти-провайдерный синтез     |
-| 🛡️ **Модерация**             | `/v1/moderations` — Проверки безопасности контента  |
-| 🔀 **Reranking**             | `/v1/rerank` — Переранжирование релевантности       |
+| Функция                      | Что делает                                                                                                                                                                 |
+| ---------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Генерация изображений** | `/v1/images/generations` — 4 провайдера, 9+ моделей                                                                                                                        |
+| 📐 **Embeddings**            | `/v1/embeddings` — 6 провайдеров, 9+ моделей                                                                                                                               |
+| 🎤 **Транскрипция аудио**    | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Текст в речь**          | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Модерация**             | `/v1/moderations` — Проверки безопасности контента                                                                                                                         |
+| 🔀 **Reranking**             | `/v1/rerank` — Переранжирование релевантности                                                                                                                              |
 
 ### 🛡️ Устойчивость и безопасность
 

docs/i18n/sk/README.md

@@ -877,14 +877,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Multimodálne API
 
-| Funkcia                     | Čo to robí                                                       |
-| --------------------------- | ---------------------------------------------------------------- |
-| 🖼️ **Generovanie obrázkov** | `/v1/images/generations` — 4 poskytovatelia, 9+ modelov          |
-| 📐 **Vloženie**             | `/v1/embeddings` — 6 poskytovateľov, 9+ modelov                  |
-| 🎤 **Prepis zvuku**         | `/v1/audio/transcriptions` — Kompatibilné so šepotom             |
-| 🔊 **Prevod textu na reč**  | `/v1/audio/speech` — Zvuková syntéza od viacerých poskytovateľov |
-| 🛡️ **Moderovania**          | `/v1/moderations` — Kontroly bezpečnosti obsahu                  |
-| 🔀 **Reranking**            | `/v1/rerank` — Zmena poradia relevantnosti dokumentu             |
+| Funkcia                     | Čo to robí                                                                                                                                                                 |
+| --------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Generovanie obrázkov** | `/v1/images/generations` — 4 poskytovatelia, 9+ modelov                                                                                                                    |
+| 📐 **Vloženie**             | `/v1/embeddings` — 6 poskytovateľov, 9+ modelov                                                                                                                            |
+| 🎤 **Prepis zvuku**         | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Prevod textu na reč**  | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderovania**          | `/v1/moderations` — Kontroly bezpečnosti obsahu                                                                                                                            |
+| 🔀 **Reranking**            | `/v1/rerank` — Zmena poradia relevantnosti dokumentu                                                                                                                       |
 
 ### 🛡️ Odolnosť a bezpečnosť
 

docs/i18n/sv/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Multimodala API:er
 
-| Funktion                 | Vad det gör                                            |
-| ------------------------ | ------------------------------------------------------ |
-| 🖼️ **Bildgenerering**    | `/v1/images/generations` — 4 leverantörer, 9+ modeller |
-| 📐 **Inbäddningar**      | `/v1/embeddings` — 6 leverantörer, 9+ modeller         |
-| 🎤 **Ljudtranskription** | `/v1/audio/transcriptions` — Whisper-kompatibel        |
-| 🔊 **Text-till-tal**     | `/v1/audio/speech` — Ljudsyntes med flera leverantörer |
-| 🛡️ **Moderationer**      | `/v1/moderations` — Innehållssäkerhetskontroller       |
-| 🔀 **Omrankning**        | `/v1/rerank` — Omrankning av dokumentrelevans          |
+| Funktion                 | Vad det gör                                                                                                                                                                |
+| ------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Bildgenerering**    | `/v1/images/generations` — 4 leverantörer, 9+ modeller                                                                                                                     |
+| 📐 **Inbäddningar**      | `/v1/embeddings` — 6 leverantörer, 9+ modeller                                                                                                                             |
+| 🎤 **Ljudtranskription** | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Text-till-tal**     | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Moderationer**      | `/v1/moderations` — Innehållssäkerhetskontroller                                                                                                                           |
+| 🔀 **Omrankning**        | `/v1/rerank` — Omrankning av dokumentrelevans                                                                                                                              |
 
 ### 🛡️ Motståndskraft och säkerhet
 

docs/i18n/th/README.md

@@ -874,14 +874,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Multi-Modal API
 
-| คุณสมบัติ               | มันทำอะไร                                                     |
-| ----------------------- | ------------------------------------------------------------- |
-| 🖼️ **การสร้างภาพ**      | `/v1/images/generations` — ผู้ให้บริการ 4 ราย รุ่น 9+         |
-| 📐 **การฝัง**           | `/v1/embeddings` — ผู้ให้บริการ 6 ราย รุ่น 9+                 |
-| 🎶 **การถอดเสียง**      | `/v1/audio/transcriptions` — รองรับการกระซิบ                  |
-| 🔊 **ข้อความเป็นคำพูด** | `/v1/audio/speech` — การสังเคราะห์เสียงจากผู้ให้บริการหลายราย |
-| 🛡️ **การกลั่นกรอง**     | `/v1/moderations` — การตรวจสอบความปลอดภัยของเนื้อหา           |
-| 🔀 **จัดอันดับ**        | `/v1/rerank` — การจัดอันดับความเกี่ยวข้องของเอกสาร            |
+| คุณสมบัติ               | มันทำอะไร                                                                                                                             |
+| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **การสร้างภาพ**      | `/v1/images/generations` — ผู้ให้บริการ 4 ราย รุ่น 9+                                                                                 |
+| 📐 **การฝัง**           | `/v1/embeddings` — ผู้ให้บริการ 6 ราย รุ่น 9+                                                                                         |
+| 🎶 **การถอดเสียง**      | `/v1/audio/transcriptions` — รองรับการกระซิบ                                                                                          |
+| 🔊 **ข้อความเป็นคำพูด** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise) |
+| 🛡️ **การกลั่นกรอง**     | `/v1/moderations` — การตรวจสอบความปลอดภัยของเนื้อหา                                                                                   |
+| 🔀 **จัดอันดับ**        | `/v1/rerank` — การจัดอันดับความเกี่ยวข้องของเอกสาร                                                                                    |
 
 ### 🛡️ ความยืดหยุ่นและความปลอดภัย
 

docs/i18n/uk-UA/README.md

@@ -878,14 +878,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 Мультимодальні API
 
-| Особливість                        | Що він робить                                         |
-| ---------------------------------- | ----------------------------------------------------- |
-| 🖼️ **Створення зображень**         | `/v1/images/generations` — 4 провайдери, 9+ моделей   |
-| 📐 **Вбудовування**                | `/v1/embeddings` — 6 провайдерів, 9+ моделей          |
-| 🎤 **Транскрипція аудіо**          | `/v1/audio/transcriptions` — сумісний із Whisper      |
-| 🔊 **Створення тексту в мовлення** | `/v1/audio/speech` — Багатопровайдерний аудіосинтез   |
-| 🛡️ **Модерації**                   | `/v1/moderations` — Перевірка безпеки вмісту          |
-| 🔀 **Переранжування**              | `/v1/rerank` — Переранжування релевантності документа |
+| Особливість                        | Що він робить                                                                                                                                                              |
+| ---------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Створення зображень**         | `/v1/images/generations` — 4 провайдери, 9+ моделей                                                                                                                        |
+| 📐 **Вбудовування**                | `/v1/embeddings` — 6 провайдерів, 9+ моделей                                                                                                                               |
+| 🎤 **Транскрипція аудіо**          | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Створення тексту в мовлення** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Модерації**                   | `/v1/moderations` — Перевірка безпеки вмісту                                                                                                                               |
+| 🔀 **Переранжування**              | `/v1/rerank` — Переранжування релевантності документа                                                                                                                      |
 
 ### 🛡️ Стійкість і безпека
 

docs/i18n/vi/README.md

@@ -874,14 +874,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 API đa phương thức
 
-| Tính năng                             | Nó làm gì                                                    |
-| ------------------------------------- | ------------------------------------------------------------ |
-| 🖼️ **Tạo hình ảnh**                   | `/v1/images/generations` — 4 nhà cung cấp, hơn 9 mô hình     |
-| 📐 **Nhúng**                          | `/v1/embeddings` — 6 nhà cung cấp, hơn 9 mô hình             |
-| 🎤 **Phiên âm âm thanh**              | `/v1/audio/transcriptions` — Tương thích với lời thì thầm    |
-| 🔊 **Chuyển văn bản thành giọng nói** | `/v1/audio/speech` — Tổng hợp âm thanh từ nhiều nhà cung cấp |
-| 🛡️ **Kiểm duyệt**                     | `/v1/moderations` — Kiểm tra an toàn nội dung                |
-| 🔀 **Sắp xếp lại**                    | `/v1/rerank` — Sắp xếp lại mức độ liên quan của tài liệu     |
+| Tính năng                             | Nó làm gì                                                                                                                                                                  |
+| ------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **Tạo hình ảnh**                   | `/v1/images/generations` — 4 nhà cung cấp, hơn 9 mô hình                                                                                                                   |
+| 📐 **Nhúng**                          | `/v1/embeddings` — 6 nhà cung cấp, hơn 9 mô hình                                                                                                                           |
+| 🎤 **Phiên âm âm thanh**              | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **Chuyển văn bản thành giọng nói** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **Kiểm duyệt**                     | `/v1/moderations` — Kiểm tra an toàn nội dung                                                                                                                              |
+| 🔀 **Sắp xếp lại**                    | `/v1/rerank` — Sắp xếp lại mức độ liên quan của tài liệu                                                                                                                   |
 
 ### 🛡️ Khả năng phục hồi và bảo mật
 

docs/i18n/zh-CN/README.md

@@ -873,14 +873,14 @@ npm run electron:build:linux   # Linux (.AppImage)
 
 ### 🎵 多模态 API
 
-| 功能              | 功能描述                                       |
-| ----------------- | ---------------------------------------------- |
-| 🖼️ **图像生成**   | `/v1/images/generations` — 4 个提供商，9+ 模型 |
-| 📐 **Embeddings** | `/v1/embeddings` — 6 个提供商，9+ 模型         |
-| 🎤 **音频转录**   | `/v1/audio/transcriptions` — Whisper 兼容      |
-| 🔊 **文字转语音** | `/v1/audio/speech` — 多提供商音频合成          |
-| 🛡️ **内容审核**   | `/v1/moderations` — 内容安全检查               |
-| 🔀 **重排序**     | `/v1/rerank` — 文档相关性重排序                |
+| 功能              | 功能描述                                                                                                                                                                   |
+| ----------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| 🖼️ **图像生成**   | `/v1/images/generations` — 4 个提供商，9+ 模型                                                                                                                             |
+| 📐 **Embeddings** | `/v1/embeddings` — 6 个提供商，9+ 模型                                                                                                                                     |
+| 🎤 **音频转录**   | `/v1/audio/transcriptions` — 7 providers (Deepgram Nova 3, AssemblyAI, Groq Whisper, HuggingFace, ElevenLabs, OpenAI, Azure), auto-language detection, MP4/MP3/WAV support |
+| 🔊 **文字转语音** | `/v1/audio/speech` — 10 providers (ElevenLabs, OpenAI, Deepgram, Cartesia, PlayHT, HuggingFace, Nvidia NIM, Inworld, Coqui, Tortoise)                                      |
+| 🛡️ **内容审核**   | `/v1/moderations` — 内容安全检查                                                                                                                                           |
+| 🔀 **重排序**     | `/v1/rerank` — 文档相关性重排序                                                                                                                                            |
 
 ### 🛡️ 弹性与安全
 

docs/openapi.yaml

@@ -1,7 +1,7 @@
 openapi: 3.1.0
 info:
   title: OmniRoute API
-  version: 2.9.0
+  version: 3.0.0-rc.7
   description: |
     OmniRoute is a local-first AI API proxy router. It provides an OpenAI-compatible
     endpoint that routes requests to multiple AI providers with load balancing,

open-sse/config/embeddingRegistry.ts

@@ -17,6 +17,7 @@ export interface EmbeddingProvider {
 }
 
 export interface EmbeddingProviderNodeRow {
+  id?: string;
   prefix: string;
   name: string;
   baseUrl: string;

open-sse/config/providerRegistry.ts

@@ -47,6 +47,8 @@ export interface RegistryEntry {
   executor: string;
   baseUrl?: string;
   baseUrls?: string[];
+  /** Override base URL used only for API key validation (e.g., opencode-go validates on zen/v1) */
+  testKeyBaseUrl?: string;
   responsesBaseUrl?: string;
   urlSuffix?: string;
   urlBuilder?: (base: string, model: string, stream: boolean) => string;
@@ -495,6 +497,41 @@ export const REGISTRY: Record<string, RegistryEntry> = {
     ],
   },
 
+  "opencode-go": {
+    id: "opencode-go",
+    alias: "opencode-go",
+    format: "openai",
+    executor: "opencode",
+    baseUrl: "https://opencode.ai/zen/go/v1",
+    // (#532) Key validation must hit the main zen endpoint (same key works for both tiers)
+    testKeyBaseUrl: "https://opencode.ai/zen/v1",
+    authType: "apikey",
+    authHeader: "Authorization",
+    authPrefix: "Bearer",
+    models: [
+      { id: "glm-5", name: "GLM-5" },
+      { id: "kimi-k2.5", name: "Kimi K2.5" },
+      { id: "minimax-m2.7", name: "MiniMax M2.7", targetFormat: "claude" },
+      { id: "minimax-m2.5", name: "MiniMax M2.5", targetFormat: "claude" },
+    ],
+  },
+
+  "opencode-zen": {
+    id: "opencode-zen",
+    alias: "opencode-zen",
+    format: "openai",
+    executor: "opencode",
+    baseUrl: "https://opencode.ai/zen/v1",
+    authType: "apikey",
+    authHeader: "Authorization",
+    authPrefix: "Bearer",
+    models: [
+      { id: "minimax-m2.5-free", name: "MiniMax M2.5 Free" },
+      { id: "big-pickle", name: "Big Pickle" },
+      { id: "gpt-5-nano", name: "GPT 5 Nano" },
+    ],
+  },
+
   openrouter: {
     id: "openrouter",
     alias: "openrouter",
@@ -883,6 +920,12 @@ export const REGISTRY: Record<string, RegistryEntry> = {
     authType: "apikey",
     authHeader: "bearer",
     models: [
+      { id: "meta-llama/Llama-3.3-70B-Instruct-Turbo-Free", name: "Llama 3.3 70B Turbo (🆓 Free)" },
+      { id: "meta-llama/Llama-Vision-Free", name: "Llama Vision (🆓 Free)" },
+      {
+        id: "deepseek-ai/DeepSeek-R1-Distill-Llama-70B-Free",
+        name: "DeepSeek R1 Distill 70B (🆓 Free)",
+      },
       { id: "meta-llama/Llama-3.3-70B-Instruct-Turbo", name: "Llama 3.3 70B Turbo" },
       { id: "deepseek-ai/DeepSeek-R1", name: "DeepSeek R1" },
       { id: "Qwen/Qwen3-235B-A22B", name: "Qwen3 235B" },
@@ -1154,6 +1197,170 @@ export const REGISTRY: Record<string, RegistryEntry> = {
     ],
     passthroughModels: true,
   },
+
+  // ── New Free Providers (2026) ─────────────────────────────────────────────
+
+  longcat: {
+    id: "longcat",
+    alias: "lc",
+    format: "openai",
+    executor: "default",
+    // (#536) Correct OpenAI-compatible base URL — was longcat.chat/api/v1/chat/completions
+    // which is the chat endpoint directly, not the base. Key validation and routing must
+    // use https://api.longcat.chat/openai which resolves /v1/models and /v1/chat/completions
+    baseUrl: "https://api.longcat.chat/openai",
+    authType: "apikey",
+    authHeader: "Authorization",
+    authPrefix: "Bearer",
+    // Free tier: 50M tokens/day (Flash-Lite) + 500K/day (Chat/Thinking) — 100% free while public beta
+    models: [
+      { id: "LongCat-Flash-Lite", name: "LongCat Flash-Lite (50M tok/day 🆓)" },
+      { id: "LongCat-Flash-Chat", name: "LongCat Flash-Chat (500K tok/day 🆓)" },
+      { id: "LongCat-Flash-Thinking", name: "LongCat Flash-Thinking (500K tok/day 🆓)" },
+      { id: "LongCat-Flash-Thinking-2601", name: "LongCat Flash-Thinking-2601 (🆓)" },
+      { id: "LongCat-Flash-Omni-2603", name: "LongCat Flash-Omni-2603 (🆓)" },
+    ],
+  },
+
+  pollinations: {
+    id: "pollinations",
+    alias: "pol",
+    format: "openai",
+    executor: "pollinations",
+    // No API key required for basic use. Proxy to GPT-5, Claude, Gemini, DeepSeek, Llama 4.
+    baseUrl: "https://text.pollinations.ai/openai/chat/completions",
+    authType: "apikey", // Optional — works without one too
+    authHeader: "bearer",
+    models: [
+      { id: "openai", name: "GPT-5 via Pollinations (🆓)" },
+      { id: "claude", name: "Claude via Pollinations (🆓)" },
+      { id: "gemini", name: "Gemini via Pollinations (🆓)" },
+      { id: "deepseek", name: "DeepSeek V3 via Pollinations (🆓)" },
+      { id: "llama", name: "Llama 4 via Pollinations (🆓)" },
+      { id: "mistral", name: "Mistral via Pollinations (🆓)" },
+    ],
+  },
+
+  puter: {
+    id: "puter",
+    alias: "pu",
+    format: "openai",
+    executor: "puter",
+    // OpenAI-compatible gateway with 500+ models (GPT, Claude, Gemini, Grok, DeepSeek, Qwen…)
+    // Auth: Bearer <puter_auth_token> from puter.com/dashboard → Copy Auth Token
+    // Model IDs use provider/model-name format for non-OpenAI models.
+    // Only chat completions (incl. streaming) are available via REST.
+    // Image gen, TTS, STT, video are puter.js SDK-only (browser).
+    baseUrl: "https://api.puter.com/puterai/openai/v1/chat/completions",
+    authType: "apikey",
+    authHeader: "bearer",
+    models: [
+      // OpenAI — use bare IDs
+      { id: "gpt-4o-mini", name: "GPT-4o Mini (🆓 Puter)" },
+      { id: "gpt-4o", name: "GPT-4o (Puter)" },
+      { id: "gpt-4.1", name: "GPT-4.1 (Puter)" },
+      { id: "gpt-4.1-mini", name: "GPT-4.1 Mini (Puter)" },
+      { id: "gpt-5-nano", name: "GPT-5 Nano (Puter)" },
+      { id: "gpt-5-mini", name: "GPT-5 Mini (Puter)" },
+      { id: "gpt-5", name: "GPT-5 (Puter)" },
+      { id: "o3-mini", name: "OpenAI o3-mini (Puter)" },
+      { id: "o3", name: "OpenAI o3 (Puter)" },
+      { id: "o4-mini", name: "OpenAI o4-mini (Puter)" },
+      // Anthropic Claude — use bare IDs (confirmed working)
+      { id: "claude-haiku-4-5", name: "Claude Haiku 4.5 (Puter)" },
+      { id: "claude-sonnet-4-5", name: "Claude Sonnet 4.5 (Puter)" },
+      { id: "claude-opus-4-5", name: "Claude Opus 4.5 (Puter)" },
+      { id: "claude-sonnet-4", name: "Claude Sonnet 4 (Puter)" },
+      { id: "claude-opus-4", name: "Claude Opus 4 (Puter)" },
+      // Google Gemini — use google/ prefix (confirmed working)
+      { id: "google/gemini-2.0-flash", name: "Gemini 2.0 Flash (Puter)" },
+      { id: "google/gemini-2.5-flash", name: "Gemini 2.5 Flash (Puter)" },
+      { id: "google/gemini-2.5-pro", name: "Gemini 2.5 Pro (Puter)" },
+      { id: "google/gemini-3-flash", name: "Gemini 3 Flash (Puter)" },
+      { id: "google/gemini-3-pro", name: "Gemini 3 Pro (Puter)" },
+      // DeepSeek — use deepseek/ prefix (confirmed working)
+      { id: "deepseek/deepseek-chat", name: "DeepSeek Chat (Puter)" },
+      { id: "deepseek/deepseek-r1", name: "DeepSeek R1 (Puter)" },
+      { id: "deepseek/deepseek-v3.2", name: "DeepSeek V3.2 (Puter)" },
+      // xAI Grok — use x-ai/ prefix
+      { id: "x-ai/grok-3", name: "Grok 3 (Puter)" },
+      { id: "x-ai/grok-3-mini", name: "Grok 3 Mini (Puter)" },
+      { id: "x-ai/grok-4", name: "Grok 4 (Puter)" },
+      { id: "x-ai/grok-4-fast", name: "Grok 4 Fast (Puter)" },
+      // Meta Llama — bare IDs (confirmed ✅)
+      { id: "llama-4-scout", name: "Llama 4 Scout (Puter)" },
+      { id: "llama-4-maverick", name: "Llama 4 Maverick (Puter)" },
+      { id: "llama-3.3-70b-instruct", name: "Llama 3.3 70B (Puter)" },
+      // Mistral — bare IDs (confirmed ✅)
+      { id: "mistral-small-latest", name: "Mistral Small (Puter)" },
+      { id: "mistral-medium-latest", name: "Mistral Medium (Puter)" },
+      { id: "open-mistral-nemo", name: "Mistral Nemo (Puter)" },
+      // Qwen — use qwen/ prefix (confirmed ✅)
+      { id: "qwen/qwen3-235b-a22b", name: "Qwen3 235B (Puter)" },
+      { id: "qwen/qwen3-32b", name: "Qwen3 32B (Puter)" },
+      { id: "qwen/qwen3-coder", name: "Qwen3 Coder 480B (Puter)" },
+    ],
+    passthroughModels: true, // 500+ models available — users can type any Puter model ID
+  },
+
+  "cloudflare-ai": {
+    id: "cloudflare-ai",
+    alias: "cf",
+    format: "openai",
+    executor: "cloudflare-ai",
+    // URL is dynamic: uses accountId from credentials. The executor builds it.
+    baseUrl: "https://api.cloudflare.com/client/v4/accounts",
+    authType: "apikey",
+    authHeader: "bearer",
+    // 10K Neurons/day free: ~150 LLM responses or 500s Whisper audio — global edge
+    models: [
+      { id: "@cf/meta/llama-3.3-70b-instruct", name: "Llama 3.3 70B (🆓 ~150 resp/day)" },
+      { id: "@cf/meta/llama-3.1-8b-instruct", name: "Llama 3.1 8B (🆓)" },
+      { id: "@cf/google/gemma-3-12b-it", name: "Gemma 3 12B (🆓)" },
+      { id: "@cf/mistral/mistral-7b-instruct-v0.2-lora", name: "Mistral 7B (🆓)" },
+      { id: "@cf/qwen/qwen2.5-coder-15b-instruct", name: "Qwen 2.5 Coder 15B (🆓)" },
+      { id: "@cf/deepseek-ai/deepseek-r1-distill-qwen-32b", name: "DeepSeek R1 Distill 32B (🆓)" },
+    ],
+  },
+
+  scaleway: {
+    id: "scaleway",
+    alias: "scw",
+    format: "openai",
+    executor: "default",
+    baseUrl: "https://api.scaleway.ai/v1/chat/completions",
+    authType: "apikey",
+    authHeader: "bearer",
+    // 1M tokens free for new accounts — EU/GDPR (Paris), no credit card needed under limit
+    models: [
+      { id: "qwen3-235b-a22b-instruct-2507", name: "Qwen3 235B A22B (1M free tok 🆓)" },
+      { id: "llama-3.1-70b-instruct", name: "Llama 3.1 70B (🆓 EU)" },
+      { id: "llama-3.1-8b-instruct", name: "Llama 3.1 8B (🆓 EU)" },
+      { id: "mistral-small-3.2-24b-instruct-2506", name: "Mistral Small 3.2 (🆓 EU)" },
+      { id: "deepseek-v3-0324", name: "DeepSeek V3 (🆓 EU)" },
+      { id: "gpt-oss-120b", name: "GPT-OSS 120B (🆓 EU)" },
+    ],
+  },
+
+  aimlapi: {
+    id: "aimlapi",
+    alias: "aiml",
+    format: "openai",
+    executor: "default",
+    baseUrl: "https://api.aimlapi.com/v1/chat/completions",
+    authType: "apikey",
+    authHeader: "bearer",
+    // $0.025/day free credits — 200+ models via single aggregator endpoint
+    models: [
+      { id: "gpt-4o", name: "GPT-4o (via AI/ML API)" },
+      { id: "claude-3-5-sonnet-20241022", name: "Claude 3.5 Sonnet (via AI/ML API)" },
+      { id: "gemini-1.5-pro", name: "Gemini 1.5 Pro (via AI/ML API)" },
+      { id: "meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo", name: "Llama 3.1 70B (via AI/ML API)" },
+      { id: "deepseek-chat", name: "DeepSeek Chat (via AI/ML API)" },
+      { id: "mistral-large-latest", name: "Mistral Large (via AI/ML API)" },
+    ],
+    passthroughModels: true,
+  },
 };
 
 // ── Generator Functions ───────────────────────────────────────────────────

open-sse/executors/antigravity.ts

@@ -44,12 +44,28 @@ export class AntigravityExecutor extends BaseExecutor {
     // stale/wrong client-side values causing 404/403 from Cloud Code endpoints.
     // Opt-in escape hatch: set OMNIROUTE_ALLOW_BODY_PROJECT_OVERRIDE=1.
     const projectId =
-      allowBodyProjectOverride && bodyProjectId ? bodyProjectId : credentialsProjectId || bodyProjectId;
+      allowBodyProjectOverride && bodyProjectId
+        ? bodyProjectId
+        : credentialsProjectId || bodyProjectId;
 
     if (!projectId) {
-      throw new Error(
-        "Missing Google projectId for Antigravity account. Please reconnect OAuth so OmniRoute can fetch your real Cloud Code project (loadCodeAssist)."
-      );
+      // (#489) Return a structured error instead of throwing — gives the client a clear signal
+      // to show a "Reconnect OAuth" prompt rather than an opaque "Internal Server Error".
+      const errorMsg =
+        "Missing Google projectId for Antigravity account. Please reconnect OAuth in Providers → Antigravity so OmniRoute can fetch your Cloud Code project.";
+      const errorBody = {
+        error: {
+          message: errorMsg,
+          type: "oauth_missing_project_id",
+          code: "missing_project_id",
+        },
+      };
+      const resp = new Response(JSON.stringify(errorBody), {
+        status: 422,
+        headers: { "Content-Type": "application/json" },
+      });
+      // Returning a Response object signals the executor to stop and forward it
+      return resp as unknown as never;
     }
 
     // Fix contents for Claude models via Antigravity

open-sse/executors/cloudflare-ai.ts

@@ -0,0 +1,59 @@
+import { BaseExecutor } from "./base.ts";
+import { PROVIDERS } from "../config/constants.ts";
+
+/**
+ * CloudflareAIExecutor — handles dynamic URL construction with accountId.
+ * Cloudflare Workers AI uses the authenticated user's account ID in the URL.
+ *
+ * URL pattern: https://api.cloudflare.com/client/v4/accounts/{accountId}/ai/v1/chat/completions
+ * Auth: Bearer <API Token>
+ * Docs: https://developers.cloudflare.com/workers-ai/
+ *
+ * Free tier: 10,000 Neurons/day = ~150 LLM responses or 500s Whisper audio
+ * API Token: dash.cloudflare.com/profile/api-tokens
+ * Account ID: right sidebar of dash.cloudflare.com
+ */
+export class CloudflareAIExecutor extends BaseExecutor {
+  constructor() {
+    super("cloudflare-ai", PROVIDERS["cloudflare-ai"] || { format: "openai" });
+  }
+
+  buildUrl(_model: string, _stream: boolean, _urlIndex = 0, credentials: any = null): string {
+    // Account ID can be stored in providerSpecificData or at top level credentials
+    const accountId =
+      credentials?.providerSpecificData?.accountId ||
+      credentials?.accountId ||
+      process.env.CLOUDFLARE_ACCOUNT_ID;
+
+    if (!accountId) {
+      throw new Error(
+        "Cloudflare Workers AI requires an Account ID. " +
+          "Add it in provider settings under 'Account ID'. " +
+          "Find it at: https://dash.cloudflare.com (right sidebar)."
+      );
+    }
+
+    return `https://api.cloudflare.com/client/v4/accounts/${accountId}/ai/v1/chat/completions`;
+  }
+
+  buildHeaders(credentials: any, stream = true): Record<string, string> {
+    const headers: Record<string, string> = {
+      "Content-Type": "application/json",
+      Authorization: `Bearer ${credentials.apiKey || credentials.accessToken}`,
+    };
+
+    if (stream) {
+      headers["Accept"] = "text/event-stream";
+    }
+
+    return headers;
+  }
+
+  transformRequest(_model: string, body: any, _stream: boolean, _credentials: any): any {
+    // Cloudflare uses full model paths like @cf/meta/llama-3.3-70b-instruct
+    // No transformation needed — user sends the full Cloudflare model path.
+    return body;
+  }
+}
+
+export default CloudflareAIExecutor;

open-sse/executors/codex.ts

@@ -3,6 +3,112 @@ import { CODEX_DEFAULT_INSTRUCTIONS } from "../config/codexInstructions.ts";
 import { PROVIDERS } from "../config/constants.ts";
 import { refreshCodexToken } from "../services/tokenRefresh.ts";
 
+// ─── T09: Codex vs Spark Scope-Aware Rate Limiting ────────────────────────
+// Codex has two independent quota pools: "codex" (standard) and "spark" (premium).
+// Exhausting one should NOT block requests to the other.
+// Ref: sub2api PR #1129 (feat(openai): split codex spark rate limiting from codex)
+
+/**
+ * Maps model name substrings to their rate-limit scope.
+ * Checked in order — first match wins.
+ */
+const CODEX_SCOPE_PATTERNS: Array<{ pattern: string; scope: "codex" | "spark" }> = [
+  { pattern: "codex-spark", scope: "spark" },
+  { pattern: "spark", scope: "spark" },
+  { pattern: "codex", scope: "codex" },
+  { pattern: "gpt-5", scope: "codex" }, // gpt-5.2-codex, gpt-5.3-codex, etc.
+];
+
+/**
+ * T09: Determine the rate-limit scope for a Codex model.
+ * Use this key as the suffix for per-scope rate limit state:
+ *   `${accountId}:${getModelScope(model)}`
+ *
+ * @param model - The Codex model ID (e.g. "gpt-5.3-codex", "codex-spark-mini")
+ * @returns "codex" | "spark"
+ */
+export function getCodexModelScope(model: string): "codex" | "spark" {
+  const lower = model.toLowerCase();
+  for (const { pattern, scope } of CODEX_SCOPE_PATTERNS) {
+    if (lower.includes(pattern)) return scope;
+  }
+  return "codex"; // default scope
+}
+
+/**
+ * T09: Get the scope-keyed rate limit identifier for an account+model combination.
+ * Use this as the key for rateLimitState maps to ensure scope isolation.
+ */
+export function getCodexRateLimitKey(accountId: string, model: string): string {
+  return `${accountId}:${getCodexModelScope(model)}`;
+}
+
+/**
+ * T03: Parsed quota snapshot from Codex response headers.
+ * Codex includes per-account usage windows that allow precise reset scheduling.
+ * Ref: sub2api PR #357 (feat(oauth): persist usage snapshots and window cooldown)
+ */
+export interface CodexQuotaSnapshot {
+  usage5h: number; // tokens used in 5h window
+  limit5h: number; // token limit for 5h window
+  resetAt5h: string | null; // ISO timestamp when 5h window resets
+  usage7d: number; // tokens used in 7d window
+  limit7d: number; // token limit for 7d window
+  resetAt7d: string | null; // ISO timestamp when 7d window resets
+}
+
+/**
+ * T03: Parse Codex-specific quota headers from a provider response.
+ * Returns null if none of the relevant headers are present.
+ *
+ * Extracts:
+ *   x-codex-5h-usage / x-codex-5h-limit / x-codex-5h-reset-at
+ *   x-codex-7d-usage / x-codex-7d-limit / x-codex-7d-reset-at
+ */
+export function parseCodexQuotaHeaders(headers: Headers): CodexQuotaSnapshot | null {
+  const usage5h = headers.get("x-codex-5h-usage");
+  const limit5h = headers.get("x-codex-5h-limit");
+  const resetAt5h = headers.get("x-codex-5h-reset-at");
+  const usage7d = headers.get("x-codex-7d-usage");
+  const limit7d = headers.get("x-codex-7d-limit");
+  const resetAt7d = headers.get("x-codex-7d-reset-at");
+
+  // Return null if none of the quota headers are present (not a quota-aware response)
+  if (!usage5h && !limit5h && !resetAt5h && !usage7d && !limit7d && !resetAt7d) {
+    return null;
+  }
+
+  return {
+    usage5h: usage5h ? parseFloat(usage5h) : 0,
+    limit5h: limit5h ? parseFloat(limit5h) : Infinity,
+    resetAt5h: resetAt5h ?? null,
+    usage7d: usage7d ? parseFloat(usage7d) : 0,
+    limit7d: limit7d ? parseFloat(limit7d) : Infinity,
+    resetAt7d: resetAt7d ?? null,
+  };
+}
+
+/**
+ * T03: Get the soonest quota reset time from a CodexQuotaSnapshot.
+ * 7d window takes priority (wider window, harder limit) but we use whichever
+ * is further in the future to avoid releasing the block too early.
+ *
+ * @returns Unix timestamp (ms) of the soonest effective reset, or null
+ */
+export function getCodexResetTime(quota: CodexQuotaSnapshot): number | null {
+  const times: number[] = [];
+  if (quota.resetAt7d) {
+    const t = new Date(quota.resetAt7d).getTime();
+    if (!isNaN(t) && t > Date.now()) times.push(t);
+  }
+  if (quota.resetAt5h) {
+    const t = new Date(quota.resetAt5h).getTime();
+    if (!isNaN(t) && t > Date.now()) times.push(t);
+  }
+  if (times.length === 0) return null;
+  return Math.max(...times); // Use furthest-out reset to avoid premature unblock
+}
+
 // Ordered list of effort levels from lowest to highest
 const EFFORT_ORDER = ["none", "low", "medium", "high", "xhigh"] as const;
 type EffortLevel = (typeof EFFORT_ORDER)[number];

open-sse/executors/default.ts

@@ -82,7 +82,7 @@ export class DefaultExecutor extends BaseExecutor {
   /**
    * For compatible providers, the model name is already clean by the time
    * it reaches the executor (chatCore sets body.model = modelInfo.model,
-   * which is the parsed model ID without any internal routing prefix).
+   * which is the parsed model ID without internal routing prefixes).
    *
    * Models may legitimately contain "/" as part of their ID (e.g. "zai-org/GLM-5-FP8",
    * "org/model-name") — we must NOT strip path segments. (Fix #493)

open-sse/executors/index.ts

@@ -6,6 +6,10 @@ import { KiroExecutor } from "./kiro.ts";
 import { CodexExecutor } from "./codex.ts";
 import { CursorExecutor } from "./cursor.ts";
 import { DefaultExecutor } from "./default.ts";
+import { PollinationsExecutor } from "./pollinations.ts";
+import { CloudflareAIExecutor } from "./cloudflare-ai.ts";
+import { OpencodeExecutor } from "./opencode.ts";
+import { PuterExecutor } from "./puter.ts";
 
 const executors = {
   antigravity: new AntigravityExecutor(),
@@ -16,6 +20,14 @@ const executors = {
   codex: new CodexExecutor(),
   cursor: new CursorExecutor(),
   cu: new CursorExecutor(), // Alias for cursor
+  pollinations: new PollinationsExecutor(),
+  pol: new PollinationsExecutor(), // Alias
+  "cloudflare-ai": new CloudflareAIExecutor(),
+  cf: new CloudflareAIExecutor(), // Alias
+  "opencode-zen": new OpencodeExecutor("opencode-zen"),
+  "opencode-go": new OpencodeExecutor("opencode-go"),
+  puter: new PuterExecutor(),
+  pu: new PuterExecutor(), // Alias
 };
 
 const defaultCache = new Map();
@@ -39,3 +51,7 @@ export { KiroExecutor } from "./kiro.ts";
 export { CodexExecutor } from "./codex.ts";
 export { CursorExecutor } from "./cursor.ts";
 export { DefaultExecutor } from "./default.ts";
+export { PollinationsExecutor } from "./pollinations.ts";
+export { CloudflareAIExecutor } from "./cloudflare-ai.ts";
+export { OpencodeExecutor } from "./opencode.ts";
+export { PuterExecutor } from "./puter.ts";

open-sse/executors/opencode.ts

@@ -0,0 +1,61 @@
+import { BaseExecutor, type ExecuteInput, type ProviderCredentials } from "./base.ts";
+import { PROVIDERS } from "../config/constants.ts";
+import { getModelTargetFormat } from "../config/providerModels.ts";
+
+export class OpencodeExecutor extends BaseExecutor {
+  _requestFormat: string | null = null;
+
+  constructor(provider: string) {
+    super(provider, PROVIDERS[provider] || PROVIDERS.openai);
+  }
+
+  async execute(input: ExecuteInput) {
+    this._requestFormat = getModelTargetFormat(this.provider, input.model) || "openai";
+    try {
+      return await super.execute(input);
+    } finally {
+      this._requestFormat = null;
+    }
+  }
+
+  buildUrl(
+    model: string,
+    stream: boolean,
+    urlIndex = 0,
+    credentials: ProviderCredentials | null = null
+  ) {
+    void urlIndex;
+    void credentials;
+
+    const base = this.config.baseUrl;
+    switch (this._requestFormat) {
+      case "claude":
+        return `${base}/messages`;
+      case "openai-responses":
+        return `${base}/responses`;
+      case "gemini":
+        return `${base}/models/${model}:${stream ? "streamGenerateContent?alt=sse" : "generateContent"}`;
+      default:
+        return `${base}/chat/completions`;
+    }
+  }
+
+  buildHeaders(credentials: ProviderCredentials | null, stream = true) {
+    const headers: Record<string, string> = { "Content-Type": "application/json" };
+    const key = credentials?.apiKey || credentials?.accessToken;
+
+    if (key) {
+      headers["Authorization"] = `Bearer ${key}`;
+    }
+
+    if (this._requestFormat === "claude") {
+      headers["anthropic-version"] = "2023-06-01";
+    }
+
+    if (stream) {
+      headers["Accept"] = "text/event-stream";
+    }
+
+    return headers;
+  }
+}

open-sse/executors/pollinations.ts

@@ -0,0 +1,46 @@
+import { BaseExecutor } from "./base.ts";
+import { PROVIDERS } from "../config/constants.ts";
+
+/**
+ * PollinationsExecutor — handles optional API key auth.
+ * Pollinations AI works WITHOUT any API key for basic use (1 req/15s).
+ * If an API key is provided, higher rate limits apply.
+ *
+ * Endpoint: https://text.pollinations.ai/openai/chat/completions
+ * Docs: https://pollinations.ai/docs
+ */
+export class PollinationsExecutor extends BaseExecutor {
+  constructor() {
+    super("pollinations", PROVIDERS["pollinations"] || { format: "openai" });
+  }
+
+  buildUrl(_model: string, _stream: boolean, _urlIndex = 0, _credentials = null): string {
+    return "https://text.pollinations.ai/openai/chat/completions";
+  }
+
+  buildHeaders(credentials: any, stream = true): Record<string, string> {
+    const headers: Record<string, string> = {
+      "Content-Type": "application/json",
+    };
+
+    // API key is OPTIONAL — skip Authorization header if no key provided
+    const key = credentials?.apiKey || credentials?.accessToken;
+    if (key) {
+      headers["Authorization"] = `Bearer ${key}`;
+    }
+
+    if (stream) {
+      headers["Accept"] = "text/event-stream";
+    }
+
+    return headers;
+  }
+
+  transformRequest(model: string, body: any, _stream: boolean, _credentials: any): any {
+    // Pollinations uses model names directly like "openai", "claude", "deepseek", etc.
+    // No transformation needed — the model name is already the Pollinations alias.
+    return body;
+  }
+}
+
+export default PollinationsExecutor;

open-sse/executors/puter.ts

@@ -0,0 +1,59 @@
+import { BaseExecutor } from "./base.ts";
+import { PROVIDERS } from "../config/constants.ts";
+
+/**
+ * PuterExecutor — OpenAI-compatible proxy for Puter AI.
+ *
+ * Puter exposes 500+ models (GPT, Claude, Gemini, Grok, DeepSeek, Qwen, Mistral...)
+ * through a single OpenAI-compatible REST endpoint.
+ *
+ * Endpoint: https://api.puter.com/puterai/openai/v1/chat/completions
+ * Auth:     Bearer <puter_auth_token>  (from puter.com/dashboard → Copy Auth Token)
+ * Docs:     https://docs.puter.com/AI/
+ *
+ * Model ID examples:
+ *   OpenAI:   "gpt-4o-mini", "gpt-4o", "gpt-4.1"
+ *   Claude:   "claude-sonnet-4-5", "claude-opus-4", "claude-haiku-4-5"
+ *   Gemini:   "google/gemini-2.0-flash", "google/gemini-2.5-pro"
+ *   DeepSeek: "deepseek/deepseek-chat", "deepseek/deepseek-r1"
+ *   Grok:     "x-ai/grok-3", "x-ai/grok-4"
+ *   Mistral:  "mistralai/mistral-small-3.2"
+ *   Meta:     "meta-llama/llama-3.3-70b-instruct"
+ *
+ * Note: Image generation, TTS, STT, and video are puter.js SDK-only features.
+ * Only text chat completions (with streaming SSE) are available via REST.
+ */
+export class PuterExecutor extends BaseExecutor {
+  constructor() {
+    super("puter", PROVIDERS["puter"] || { format: "openai" });
+  }
+
+  buildUrl(_model: string, _stream: boolean, _urlIndex = 0, _credentials = null): string {
+    return "https://api.puter.com/puterai/openai/v1/chat/completions";
+  }
+
+  buildHeaders(credentials: any, stream = true): Record<string, string> {
+    const headers: Record<string, string> = {
+      "Content-Type": "application/json",
+    };
+
+    const key = credentials?.apiKey || credentials?.accessToken;
+    if (key) {
+      headers["Authorization"] = `Bearer ${key}`;
+    }
+
+    if (stream) {
+      headers["Accept"] = "text/event-stream";
+    }
+
+    return headers;
+  }
+
+  transformRequest(model: string, body: any, _stream: boolean, _credentials: any): any {
+    // Puter accepts model IDs directly from its catalog.
+    // No transformation required — model string is passed as-is.
+    return body;
+  }
+}
+
+export default PuterExecutor;

open-sse/handlers/audioSpeech.ts

@@ -28,13 +28,17 @@ function upstreamErrorResponse(res, errText) {
   let errorMessage: string;
   try {
     const parsed = JSON.parse(errText);
-    errorMessage =
+    // Extract a human-readable message from various error response shapes.
+    // Guard against `parsed.error` being an object (e.g. ElevenLabs returns
+    // { error: { message: "...", status_code: 401 } } or { detail: { ... } })
+    const raw =
       parsed?.err_msg ||
       parsed?.error?.message ||
-      parsed?.error ||
+      (typeof parsed?.error === "string" ? parsed.error : null) ||
       parsed?.message ||
-      parsed?.detail ||
-      errText;
+      (typeof parsed?.detail === "string" ? parsed.detail : parsed?.detail?.message) ||
+      null;
+    errorMessage = raw ? String(raw) : errText || `Upstream error (${res.status})`;
   } catch {
     errorMessage = errText || `Upstream error (${res.status})`;
   }

open-sse/handlers/audioTranscription.ts

@@ -34,13 +34,15 @@ function upstreamErrorResponse(res, errText) {
   let errorMessage: string;
   try {
     const parsed = JSON.parse(errText);
-    errorMessage =
+    // Guard against `parsed.error` or `parsed.detail` being objects
+    const raw =
       parsed?.err_msg ||
       parsed?.error?.message ||
-      parsed?.error ||
+      (typeof parsed?.error === "string" ? parsed.error : null) ||
       parsed?.message ||
-      parsed?.detail ||
-      errText;
+      (typeof parsed?.detail === "string" ? parsed.detail : parsed?.detail?.message) ||
+      null;
+    errorMessage = raw ? String(raw) : errText || `Upstream error (${res.status})`;
   } catch {
     errorMessage = errText || `Upstream error (${res.status})`;
   }
@@ -65,13 +67,67 @@ function getUploadedFileName(file: Blob & { name?: unknown }): string {
   return typeof file.name === "string" && file.name.length > 0 ? file.name : "audio.wav";
 }
 
+/**
+ * Infer a suitable Content-Type for Deepgram from the browser-provided MIME
+ * type and the original filename.  Deepgram accepts `audio/*` and many raw
+ * formats, but `video/*` causes it to silently fail with "no speech detected".
+ *
+ * Strategy:
+ * 1. If the browser says `audio/*`, keep it as-is.
+ * 2. If it's `video/*` (e.g. `.mp4`), remap to the audio equivalent so
+ *    Deepgram extracts the audio track.  `.mp4` → `audio/mp4`, etc.
+ * 3. Fall back to `application/octet-stream` which tells Deepgram to
+ *    auto-detect from the raw bytes (most reliable for unknown formats).
+ */
+function resolveAudioContentType(file: Blob & { name?: unknown }): string {
+  const browserType = (file.type || "").toLowerCase();
+  const fileName = typeof file.name === "string" ? file.name.toLowerCase() : "";
+
+  // 1) Browser already says it's audio — trust it
+  if (browserType.startsWith("audio/")) return browserType;
+
+  // 2) Derive from file extension (covers video/* and empty MIME)
+  const ext = fileName.includes(".") ? fileName.split(".").pop() : "";
+  const EXT_TO_MIME: Record<string, string> = {
+    mp3: "audio/mpeg",
+    mp4: "audio/mp4",
+    m4a: "audio/mp4",
+    wav: "audio/wav",
+    ogg: "audio/ogg",
+    flac: "audio/flac",
+    webm: "audio/webm",
+    aac: "audio/aac",
+    wma: "audio/x-ms-wma",
+    opus: "audio/opus",
+  };
+  if (ext && EXT_TO_MIME[ext]) return EXT_TO_MIME[ext];
+
+  // 3) Fallback — let Deepgram auto-detect from raw bytes
+  return "application/octet-stream";
+}
+
 /**
  * Handle Deepgram transcription (raw binary audio, model via query param)
  */
-async function handleDeepgramTranscription(providerConfig, file, modelId, token) {
+async function handleDeepgramTranscription(
+  providerConfig,
+  file,
+  modelId,
+  token,
+  formData?: FormData
+) {
   const url = new URL(providerConfig.baseUrl);
   url.searchParams.set("model", modelId);
   url.searchParams.set("smart_format", "true");
+  url.searchParams.set("punctuate", "true");
+
+  // Language: if caller specified one, use it; otherwise let Deepgram auto-detect
+  const langParam = formData?.get("language");
+  if (typeof langParam === "string" && langParam.trim()) {
+    url.searchParams.set("language", langParam.trim());
+  } else {
+    url.searchParams.set("detect_language", "true");
+  }
 
   const arrayBuffer = await file.arrayBuffer();
 
@@ -79,7 +135,7 @@ async function handleDeepgramTranscription(providerConfig, file, modelId, token)
     method: "POST",
     headers: {
       ...buildAuthHeaders(providerConfig, token),
-      "Content-Type": file.type || "audio/wav",
+      "Content-Type": resolveAudioContentType(file),
     },
     body: arrayBuffer,
   });
@@ -212,7 +268,7 @@ async function handleHuggingFaceTranscription(providerConfig, file, modelId, tok
     method: "POST",
     headers: {
       ...buildAuthHeaders(providerConfig, token),
-      "Content-Type": file.type || "audio/wav",
+      "Content-Type": resolveAudioContentType(file),
     },
     body: arrayBuffer,
   });
@@ -283,7 +339,7 @@ export async function handleAudioTranscription({
 
   // Route to provider-specific handler
   if (providerConfig.format === "deepgram") {
-    return handleDeepgramTranscription(providerConfig, file, modelId, token);
+    return handleDeepgramTranscription(providerConfig, file, modelId, token, formData);
   }
 
   if (providerConfig.format === "assemblyai") {

open-sse/handlers/chatCore.ts

@@ -308,6 +308,27 @@ export async function handleChatCore({
                   }
                   return [];
                 }
+                // (#527) tool_result → convert to text instead of dropping.
+                // When Claude Code + superpowers routes through Codex, it sends tool_result
+                // blocks in user messages. Silently dropping them causes Codex to loop
+                // because it never receives the tool response and keeps re-requesting it.
+                if (block.type === "tool_result") {
+                  const toolId = block.tool_use_id ?? block.id ?? "unknown";
+                  const resultContent = block.content ?? block.text ?? block.output ?? "";
+                  const resultText =
+                    typeof resultContent === "string"
+                      ? resultContent
+                      : Array.isArray(resultContent)
+                        ? resultContent
+                            .filter((c: Record<string, unknown>) => c.type === "text")
+                            .map((c: Record<string, unknown>) => c.text)
+                            .join("\n")
+                        : JSON.stringify(resultContent);
+                  if (resultText.length > 0) {
+                    return [{ type: "text", text: `[Tool Result: ${toolId}]\n${resultText}` }];
+                  }
+                  return [];
+                }
                 // Unknown types: drop silently
                 log?.debug?.("CONTENT", `Dropped unsupported content part type="${block.type}"`);
                 return [];
@@ -317,10 +338,15 @@ export async function handleChatCore({
         }
       }
 
-      const normalizeToolCallId = getModelNormalizeToolCallId(provider || "", model || "");
+      const normalizeToolCallId = getModelNormalizeToolCallId(
+        provider || "",
+        model || "",
+        sourceFormat
+      );
       const preserveDeveloperRole = getModelPreserveOpenAIDeveloperRole(
         provider || "",
-        model || ""
+        model || "",
+        sourceFormat
       );
       translatedBody = translateRequest(
         sourceFormat,

open-sse/services/accountFallback.ts

@@ -8,6 +8,46 @@ import {
 } from "../config/constants.ts";
 import { getProviderCategory } from "../config/providerRegistry.ts";
 
+// T06 (sub2api PR #1037): Signals that indicate permanent account deactivation.
+// When a 401 body contains these strings, the account is permanently dead
+// and should NOT be retried after token refresh.
+export const ACCOUNT_DEACTIVATED_SIGNALS = [
+  "account_deactivated",
+  "account has been deactivated",
+  "account has been disabled",
+  "your account has been suspended",
+  "this account is deactivated",
+];
+
+// T10 (sub2api PR #1169): Signals that indicate billing credits are exhausted.
+// Distinct from rate-limit 429 — the account won't recover until credits are added.
+export const CREDITS_EXHAUSTED_SIGNALS = [
+  "insufficient_quota",
+  "billing_hard_limit_reached",
+  "exceeded your current quota",
+  "credit_balance_too_low",
+  "your credit balance is too low",
+  "credits exhausted",
+  "out of credits",
+  "payment required",
+];
+
+/**
+ * T06: Returns true if response body indicates the account is permanently deactivated.
+ */
+export function isAccountDeactivated(errorText: string): boolean {
+  const lower = String(errorText || "").toLowerCase();
+  return ACCOUNT_DEACTIVATED_SIGNALS.some((sig) => lower.includes(sig));
+}
+
+/**
+ * T10: Returns true if response body indicates credits/quota are permanently exhausted.
+ */
+export function isCreditsExhausted(errorText: string): boolean {
+  const lower = String(errorText || "").toLowerCase();
+  return CREDITS_EXHAUSTED_SIGNALS.some((sig) => lower.includes(sig));
+}
+
 // ─── Provider Profile Helper ────────────────────────────────────────────────
 
 /**
@@ -201,6 +241,14 @@ export function classifyErrorText(errorText) {
   ) {
     return RateLimitReason.QUOTA_EXHAUSTED;
   }
+  // T10: credits_exhausted signals
+  if (isCreditsExhausted(errorText)) {
+    return RateLimitReason.QUOTA_EXHAUSTED;
+  }
+  // T06: account_deactivated signals
+  if (isAccountDeactivated(errorText)) {
+    return RateLimitReason.AUTH_ERROR;
+  }
   if (
     lower.includes("rate limit") ||
     lower.includes("too many requests") ||
@@ -301,6 +349,26 @@ export function checkFallbackError(
     const errorStr = typeof errorText === "string" ? errorText : JSON.stringify(errorText);
     const lowerError = errorStr.toLowerCase();
 
+    // T06 (sub2api #1037): Permanent account deactivation — do NOT retry, mark as permanent failure
+    if (isAccountDeactivated(errorStr)) {
+      return {
+        shouldFallback: true,
+        cooldownMs: 365 * 24 * 60 * 60 * 1000, // 1 year = effectively permanent
+        reason: RateLimitReason.AUTH_ERROR,
+        permanent: true,
+      };
+    }
+
+    // T10 (sub2api #1169): Credits/quota exhausted — long cooldown, distinct from rate limit
+    if (isCreditsExhausted(errorStr)) {
+      return {
+        shouldFallback: true,
+        cooldownMs: COOLDOWN_MS.paymentRequired ?? 3600 * 1000, // 1h cooldown
+        reason: RateLimitReason.QUOTA_EXHAUSTED,
+        creditsExhausted: true,
+      };
+    }
+
     if (lowerError.includes("no credentials")) {
       return {
         shouldFallback: true,

open-sse/services/combo.ts

@@ -467,50 +467,47 @@ export async function handleComboChat({
           return res;
         }
 
-        // Streaming (Fix #490): append omniModel tag as a final SSE content delta
-        // before the [DONE] marker using TransformStream for zero-copy passthrough
+        // Streaming (Fix #490 + #511): prepend omniModel tag into the first
+        // non-empty content chunk so it arrives BEFORE finish_reason:stop.
+        // SDKs close the connection on finish_reason, so anything sent after
+        // that marker is silently dropped.
         if (!res.body) return res;
-        const tagContent = `\n<omniModel>${modelStr}</omniModel>`;
+        const tagContent = `\\n<omniModel>${modelStr}</omniModel>\\n`;
         const encoder = new TextEncoder();
         const decoder = new TextDecoder();
-        let buffer = "";
+        let tagInjected = false;
 
         const transform = new TransformStream({
           transform(chunk, controller) {
-            // Decode chunk and check for [DONE] marker
-            const text = decoder.decode(chunk, { stream: true });
-            buffer += text;
-
-            // Check if buffer contains the [DONE] marker
-            const doneIdx = buffer.indexOf("data: [DONE]");
-            if (doneIdx === -1) {
-              // No [DONE] yet — flush buffer as-is (keep passthrough latency low)
-              controller.enqueue(encoder.encode(buffer));
-              buffer = "";
+            if (tagInjected) {
+              // Already injected — passthrough
+              controller.enqueue(chunk);
               return;
             }
 
-            // Found [DONE] — inject tag content delta before it
-            const beforeDone = buffer.slice(0, doneIdx);
-            const afterDone = buffer.slice(doneIdx);
+            const text = decoder.decode(chunk, { stream: true });
 
-            // Build a synthetic SSE content delta chunk with the tag
-            const tagChunk = `data: ${JSON.stringify({
-              choices: [
-                {
-                  delta: { content: tagContent },
-                  index: 0,
-                  finish_reason: null,
-                },
-              ],
-            })}\n\n`;
+            // Look for the first SSE data line with non-empty content
+            // Pattern: "content":"<non-empty>" — we inject tag at the start
+            const contentMatch = text.match(/"content":"([^"]+)/);
+            if (contentMatch) {
+              // Inject tag at the beginning of the first content value
+              const injected = text.replace(
+                /"content":"([^"]+)/,
+                `"content":"${tagContent.replace(/"/g, '\\"')}$1`
+              );
+              tagInjected = true;
+              controller.enqueue(encoder.encode(injected));
+              return;
+            }
 
-            controller.enqueue(encoder.encode(beforeDone + tagChunk + afterDone));
-            buffer = "";
+            // No content yet — passthrough
+            controller.enqueue(chunk);
           },
           flush(controller) {
-            // If stream ends without [DONE], flush remaining buffer + tag
-            if (buffer.length > 0) {
+            // If stream ends without ever finding content (edge case),
+            // inject tag as a standalone chunk before the stream closes
+            if (!tagInjected) {
               const tagChunk = `data: ${JSON.stringify({
                 choices: [
                   {
@@ -520,7 +517,7 @@ export async function handleComboChat({
                   },
                 ],
               })}\n\n`;
-              controller.enqueue(encoder.encode(buffer + tagChunk));
+              controller.enqueue(encoder.encode(tagChunk));
             }
           },
         });

open-sse/services/comboAgentMiddleware.ts

@@ -34,7 +34,11 @@ interface Message {
 
 // ── Context Caching Tag ─────────────────────────────────────────────────────
 
-const CACHE_TAG_PATTERN = /<omniModel>([^<]+)<\/omniModel>/;
+// Handles both actual newlines (U+000A) and literal \n sequences injected
+// by combo.ts streaming around the <omniModel> tag (#531). Non-global so that
+// .exec() and .test() stay stateless; callers that need full replacement use
+// String.prototype.replace() which replaces all non-overlapping matches.
+const CACHE_TAG_PATTERN = /(?:\\n|\n)?<omniModel>([^<]+)<\/omniModel>(?:\\n|\n)?/;
 
 /**
  * Inject the model tag into the last assistant message (or append a new one).
@@ -165,7 +169,11 @@ export function applyComboAgentMiddleware(
   if (comboConfig.context_cache_protection) {
     pinnedModel = extractPinnedModel(messages);
     if (pinnedModel) {
-      // Model is pinned — caller should override model selection
+      // (#535) Model is pinned via <omniModel> tag — override body.model so the combo
+      // router uses exactly this model instead of picking a different one. Without this,
+      // the extracted pinnedModel is returned but body.model is unchanged, breaking
+      // context cache sessions by sending subsequent turns to a different model.
+      body = { ...body, model: pinnedModel };
     }
   }
 

open-sse/services/sessionManager.ts

@@ -173,6 +173,95 @@ export function getActiveSessions(): Array<SessionEntry & { sessionId: string; a
  */
 export function clearSessions(): void {
   sessions.clear();
+  activeSessionsByKey.clear();
+}
+
+// ─── T08: Per-API-Key Session Limit ─────────────────────────────────────────
+// Tracks concurrent sticky sessions per API key and enforces max_sessions limits.
+// Ref: sub2api PR #634 (fix: stabilize session hash + add user-level session limit)
+
+// Map: apiKeyId → Set<sessionId>
+const activeSessionsByKey = new Map<string, Set<string>>();
+
+/**
+ * T08: Get the number of currently active sessions for an API key.
+ * @param apiKeyId - The API key's UUID from the database
+ */
+export function getActiveSessionCountForKey(apiKeyId: string): number {
+  return activeSessionsByKey.get(apiKeyId)?.size ?? 0;
+}
+
+/**
+ * T08: Register a session as belonging to an API key.
+ * Call this after session creation is allowed (i.e., limit check passed).
+ */
+export function registerKeySession(apiKeyId: string, sessionId: string): void {
+  if (!activeSessionsByKey.has(apiKeyId)) {
+    activeSessionsByKey.set(apiKeyId, new Set());
+  }
+  activeSessionsByKey.get(apiKeyId)!.add(sessionId);
+}
+
+/**
+ * T08: Unregister a session from an API key's active set.
+ * Call this when the request closes or the session TTL expires.
+ */
+export function unregisterKeySession(apiKeyId: string, sessionId: string): void {
+  activeSessionsByKey.get(apiKeyId)?.delete(sessionId);
+  // Clean up empty sets to avoid memory leaks
+  if (activeSessionsByKey.get(apiKeyId)?.size === 0) {
+    activeSessionsByKey.delete(apiKeyId);
+  }
+}
+
+/**
+ * T08: Check whether adding a new session would exceed the key's max_sessions limit.
+ * Returns null if allowed, or an error object to return as a 429 response.
+ *
+ * @param apiKeyId - The API key's UUID
+ * @param maxSessions - The limit from the DB (0 = unlimited)
+ */
+export function checkSessionLimit(
+  apiKeyId: string,
+  maxSessions: number
+): { code: "SESSION_LIMIT_EXCEEDED"; message: string; limit: number; current: number } | null {
+  if (!maxSessions || maxSessions <= 0) return null; // unlimited
+  const current = getActiveSessionCountForKey(apiKeyId);
+  if (current < maxSessions) return null;
+  return {
+    code: "SESSION_LIMIT_EXCEEDED",
+    message:
+      `You have reached the maximum number of active sessions (${maxSessions}). ` +
+      `Please close unused sessions or wait for them to expire.`,
+    limit: maxSessions,
+    current,
+  };
+}
+
+/**
+ * T04: Extract an external session ID from request headers.
+ * Accepts both hyphenated and underscore forms for Nginx compatibility.
+ * Nginx drops headers with underscores by default — use `underscores_in_headers on`
+ * in nginx.conf, or use X-Session-Id (hyphenated) which passes cleanly.
+ *
+ * Ref: sub2api README + PR #634
+ *
+ * @param headers - Request headers (Headers object or plain object with .get())
+ * @returns External session ID with "ext:" prefix, or null
+ */
+export function extractExternalSessionId(
+  headers: Headers | { get?: (n: string) => string | null } | null | undefined
+): string | null {
+  if (!headers || typeof (headers as Headers).get !== "function") return null;
+  const h = headers as Headers;
+  const raw =
+    h.get("x-session-id") ?? // Preferred: hyphenated (passes through Nginx)
+    h.get("x-omniroute-session") ?? // OmniRoute-specific form
+    h.get("session-id") ?? // Bare session-id
+    null;
+  if (!raw || !raw.trim()) return null;
+  // Prefix "ext:" to ensure no collision with internal SHA-256 hash IDs
+  return `ext:${raw.trim().slice(0, 64)}`; // max 64 chars to avoid abuse
 }
 
 // ─── Internal Helpers ───────────────────────────────────────────────────────

open-sse/services/thinkingBudget.ts

@@ -19,6 +19,8 @@ export const EFFORT_BUDGETS = {
   low: 1024,
   medium: 10240,
   high: 131072,
+  max: 131072, // T11: Claude "max" / "xhigh" — full budget
+  xhigh: 131072, // T11: explicit alias used internally
 };
 
 // thinkingLevel string → budget token mapping
@@ -28,6 +30,8 @@ export const THINKING_LEVEL_MAP = {
   low: 1024,
   medium: 10240,
   high: 131072,
+  max: 131072, // T11: max = full Claude budget (sub2api: xhigh)
+  xhigh: 131072, // T11: explicit xhigh alias
 };
 
 // Default config (passthrough = backward compatible)
@@ -198,7 +202,7 @@ function setCustomBudget(body, budget) {
     };
   }
 
-  // OpenAI reasoning_effort mapping
+  // OpenAI reasoning_effort mapping (T11: add 'max' tier for full budget)
   if (result.reasoning_effort !== undefined || result.reasoning !== undefined) {
     if (budget <= 0) {
       delete result.reasoning_effort;
@@ -207,8 +211,10 @@ function setCustomBudget(body, budget) {
       result.reasoning_effort = "low";
     } else if (budget <= 10240) {
       result.reasoning_effort = "medium";
-    } else {
+    } else if (budget < 131072) {
       result.reasoning_effort = "high";
+    } else {
+      result.reasoning_effort = "max"; // T11: full budget → "max"
     }
   }
 

open-sse/translator/helpers/responsesApiHelper.ts

@@ -93,10 +93,11 @@ export function convertResponsesApiFormat(body) {
   }
 
   // Cleanup Responses API specific fields
+  // Note: prompt_cache_key is intentionally preserved — it is used by Codex and other
+  // providers as a cache-affinity signal. Stripping it breaks prompt caching (#517).
   delete result.input;
   delete result.instructions;
   delete result.include;
-  delete result.prompt_cache_key;
   delete result.store;
   delete result.reasoning;
 

open-sse/translator/request/openai-responses.ts

@@ -227,10 +227,11 @@ export function openaiResponsesToOpenAIRequest(
   });
 
   // Cleanup Responses API specific fields
+  // Note: prompt_cache_key is intentionally preserved — it is used by Codex and other
+  // providers as a cache-affinity signal. Stripping it breaks prompt caching (#517).
   delete result.input;
   delete result.instructions;
   delete result.include;
-  delete result.prompt_cache_key;
   delete result.store;
   delete result.reasoning;
 

open-sse/translator/request/openai-to-claude.ts

@@ -27,6 +27,60 @@ type ClaudeTool = {
   defer_loading?: boolean;
 };
 
+/**
+ * T02: Recursively strips empty text blocks from content arrays.
+ * Anthropic returns 400 "text content blocks must be non-empty" if any
+ * text block has text: "". Must also recurse into nested tool_result.content.
+ * Ref: sub2api PR #1212
+ */
+export function stripEmptyTextBlocks(content: unknown[] | undefined): unknown[] {
+  if (!Array.isArray(content)) return content ?? [];
+  return content
+    .filter((block: unknown) => {
+      if (
+        block &&
+        typeof block === "object" &&
+        (block as Record<string, unknown>).type === "text"
+      ) {
+        const text = (block as Record<string, unknown>).text;
+        if (text === "" || text == null) return false;
+      }
+      return true;
+    })
+    .map((block: unknown) => {
+      if (
+        block &&
+        typeof block === "object" &&
+        (block as Record<string, unknown>).type === "tool_result" &&
+        Array.isArray((block as Record<string, unknown>).content)
+      ) {
+        // Recurse into nested tool_result.content
+        return {
+          ...(block as Record<string, unknown>),
+          content: stripEmptyTextBlocks((block as Record<string, unknown>).content as unknown[]),
+        };
+      }
+      return block;
+    });
+}
+
+/**
+ * T15: Normalize content to string form.
+ * Handles both string and array-of-blocks forms (Cursor, Codex 2.x, etc.).
+ * Ref: sub2api PR #1197
+ */
+export function normalizeContentToString(content: string | unknown[] | null | undefined): string {
+  if (!content) return "";
+  if (typeof content === "string") return content;
+  if (Array.isArray(content)) {
+    return (content as Array<Record<string, unknown>>)
+      .filter((b) => b.type === "text")
+      .map((b) => String(b.text ?? ""))
+      .join("\n");
+  }
+  return "";
+}
+
 // Convert OpenAI request to Claude format
 export function openaiToClaudeRequest(model, body, stream) {
   // Check if tool prefix should be disabled (configured per-provider or global)
@@ -61,11 +115,11 @@ export function openaiToClaudeRequest(model, body, stream) {
   const systemParts = [];
 
   if (body.messages && Array.isArray(body.messages)) {
-    // Extract system messages
+    // Extract system messages (T15: handle both string and array content)
     for (const msg of body.messages) {
       if (msg.role === "system") {
         systemParts.push(
-          typeof msg.content === "string" ? msg.content : extractTextContent(msg.content)
+          typeof msg.content === "string" ? msg.content : normalizeContentToString(msg.content)
         );
       }
     }
@@ -270,10 +324,14 @@ function getContentBlocksFromMessage(msg, toolNameMap = new Map(), disableToolPr
   const blocks = [];
 
   if (msg.role === "tool") {
+    // T02: Strip empty text blocks from nested tool_result content to avoid Anthropic 400
+    const toolContent = Array.isArray(msg.content)
+      ? stripEmptyTextBlocks(msg.content)
+      : msg.content;
     blocks.push({
       type: "tool_result",
       tool_use_id: msg.tool_call_id,
-      content: msg.content,
+      content: toolContent,
     });
   } else if (msg.role === "user") {
     if (typeof msg.content === "string") {
@@ -287,10 +345,14 @@ function getContentBlocksFromMessage(msg, toolNameMap = new Map(), disableToolPr
         } else if (part.type === "tool_result") {
           // Skip tool_result with no tool_use_id (would be useless and may cause errors)
           if (!part.tool_use_id) continue;
+          // T02: strip empty text blocks from nested content before passing to Anthropic
+          const resultContent = Array.isArray(part.content)
+            ? stripEmptyTextBlocks(part.content)
+            : part.content;
           blocks.push({
             type: "tool_result",
             tool_use_id: part.tool_use_id,
-            content: part.content,
+            content: resultContent,
             ...(part.is_error && { is_error: part.is_error }),
           });
         } else if (part.type === "image_url") {

package.json

@@ -1,6 +1,6 @@
 {
   "name": "omniroute",
-  "version": "2.9.0",
+  "version": "3.0.0-rc.7",
   "description": "Smart AI Router with auto fallback — route to FREE & cheap models, zero downtime. Works with Cursor, Cline, Claude Desktop, Codex, and any OpenAI-compatible tool.",
   "type": "module",
   "bin": {
@@ -81,8 +81,10 @@
     "system-info": "node scripts/system-info.mjs"
   },
   "dependencies": {
+    "@lobehub/icons": "^5.0.1",
     "@modelcontextprotocol/sdk": "^1.27.1",
     "@monaco-editor/react": "^4.7.0",
+    "@swc/helpers": "0.5.19",
     "bcryptjs": "^3.0.3",
     "better-sqlite3": "^12.6.2",
     "bottleneck": "^2.19.5",
@@ -110,8 +112,7 @@
     "uuid": "^13.0.0",
     "wreq-js": "^2.0.1",
     "zod": "^4.3.6",
-    "zustand": "^5.0.10",
-    "@swc/helpers": "0.5.19"
+    "zustand": "^5.0.10"
   },
   "devDependencies": {
     "@playwright/test": "^1.58.2",

public/providers/cloudflare-ai.svg

@@ -0,0 +1 @@
+<svg role="img" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><title>Cloudflare</title><path d="M16.5088 16.8447c.1475-.5068.0908-.9707-.1553-1.3154-.2246-.3164-.6045-.499-1.0615-.5205l-8.6592-.1123a.1559.1559 0 0 1-.1333-.0713c-.0283-.042-.0351-.0986-.021-.1553.0278-.084.1123-.1484.2036-.1562l8.7359-.1123c1.0351-.0489 2.1601-.8868 2.5537-1.9136l.499-1.3013c.0215-.0561.0293-.1128.0147-.168-.5625-2.5463-2.835-4.4453-5.5499-4.4453-2.5039 0-4.6284 1.6177-5.3876 3.8614-.4927-.3658-1.1187-.5625-1.794-.499-1.2026.119-2.1665 1.083-2.2861 2.2856-.0283.31-.0069.6128.0635.894C1.5683 13.171 0 14.7754 0 16.752c0 .1748.0142.3515.0352.5273.0141.083.0844.1475.1689.1475h15.9814c.0909 0 .1758-.0645.2032-.1553l.12-.4268zm2.7568-5.5634c-.0771 0-.1611 0-.2383.0112-.0566 0-.1054.0415-.127.0976l-.3378 1.1744c-.1475.5068-.0918.9707.1543 1.3164.2256.3164.6055.498 1.0625.5195l1.8437.1133c.0557 0 .1055.0263.1329.0703.0283.043.0351.1074.0214.1562-.0283.084-.1132.1485-.204.1553l-1.921.1123c-1.041.0488-2.1582.8867-2.5527 1.914l-.1406.3585c-.0283.0713.0215.1416.0986.1416h6.5977c.0771 0 .1474-.0489.169-.126.1122-.4082.1757-.837.1757-1.2803 0-2.6025-2.125-4.727-4.7344-4.727"/></svg>

public/providers/scaleway.svg

@@ -0,0 +1 @@
+<svg role="img" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><title>Scaleway</title><path d="M16.605 11.11v5.72a1.77 1.77 0 01-1.54 1.69h-4a1.43 1.43 0 01-1.31-1.22 1.09 1.09 0 010-.18 1.37 1.37 0 011.37-1.36h1.74a1 1 0 001-1v-3.62a1.4 1.4 0 011.18-1.39h.17a1.37 1.37 0 011.39 1.36zm-6.46 1.74V9.26a1 1 0 011-1h1.85a1.37 1.37 0 001.37-1.37 1 1 0 000-.17 1.45 1.45 0 00-1.41-1.2h-3.96a1.81 1.81 0 00-1.58 1.66v5.7a1.37 1.37 0 001.37 1.37h.21a1.4 1.4 0 001.15-1.4zm12-4.29V20a4.53 4.53 0 01-4.15 4h-7.58a8.57 8.57 0 01-8.56-8.57V4.54A4.54 4.54 0 016.395 0h7.18a8.56 8.56 0 018.56 8.56zm-2.74 0a5.83 5.83 0 00-5.82-5.82h-7.19a1.79 1.79 0 00-1.8 1.8v10.89a5.83 5.83 0 005.82 5.8h7.44a1.79 1.79 0 001.54-1.48z"/></svg>

scripts/check-docs-sync.mjs

@@ -48,7 +48,8 @@ function extractChangelogSections(content) {
 }
 
 function isSemver(value) {
-  return /^\d+\.\d+\.\d+$/.test(value);
+  // Accept X.Y.Z and X.Y.Z-prerelease.N (e.g. 3.0.0-rc.1, 3.0.0-beta.2)
+  return /^\d+\.\d+\.\d+(-[a-zA-Z0-9.]+)?$/.test(value);
 }
 
 let hasFailure = false;

src/app/(dashboard)/dashboard/api-manager/ApiManagerPageClient.tsx

@@ -523,15 +523,12 @@ export default function ApiManagerPageClient() {
                   </div>
                   <div className="col-span-3 flex items-center gap-1.5">
                     <code className="text-sm text-text-muted font-mono truncate">{key.key}</code>
-                    <button
-                      onClick={() => copy(key.key, key.id)}
-                      className="p-1 hover:bg-black/5 dark:hover:bg-white/5 rounded text-text-muted hover:text-primary opacity-0 group-hover:opacity-100 transition-all shrink-0"
-                      title={t("copyMaskedKey")}
+                    <span
+                      className="p-1 text-text-muted/40 opacity-0 group-hover:opacity-100 transition-all shrink-0 cursor-help"
+                      title={t("keyOnlyAvailableAtCreation")}
                     >
-                      <span className="material-symbols-outlined text-[14px]">
-                        {copied === key.id ? "check" : "content_copy"}
-                      </span>
-                    </button>
+                      <span className="material-symbols-outlined text-[14px]">lock</span>
+                    </span>
                   </div>
                   <div className="col-span-2 flex items-center">
                     <div className="flex flex-col items-start gap-1">

src/app/(dashboard)/dashboard/cli-tools/components/ClaudeToolCard.tsx

@@ -58,8 +58,10 @@ export default function ClaudeToolCard({
   const effectiveConfigStatus = configStatus || batchStatus?.configStatus || null;
 
   useEffect(() => {
+    // (#523) Store the key *id* (not the masked string) so the backend can
+    // resolve the real secret from DB before writing to settings.json.
     if (apiKeys?.length > 0 && !selectedApiKey) {
-      setSelectedApiKey(apiKeys[0].key);
+      setSelectedApiKey(apiKeys[0].id);
     }
   }, [apiKeys, selectedApiKey]);
 
@@ -95,10 +97,11 @@ export default function ClaudeToolCard({
           }
         }
       });
-      // Only set selectedApiKey if it exists in apiKeys list
+      // Restore selected key from file: match token stored in file against known keys
       const tokenFromFile = env.ANTHROPIC_AUTH_TOKEN;
-      if (tokenFromFile && apiKeys?.some((k) => k.key === tokenFromFile)) {
-        setSelectedApiKey(tokenFromFile);
+      if (tokenFromFile) {
+        const matchedKey = apiKeys?.find((k) => k.key === tokenFromFile);
+        if (matchedKey) setSelectedApiKey(matchedKey.id);
       }
     }
   }, [claudeStatus, apiKeys, tool.defaultModels, onModelMappingChange]);
@@ -132,24 +135,27 @@ export default function ClaudeToolCard({
     try {
       const env: any = { ANTHROPIC_BASE_URL: getEffectiveBaseUrl() };
 
-      // Get key from dropdown, fallback to first key or sk_omniroute for localhost
-      const keyToUse =
-        selectedApiKey?.trim() ||
-        (apiKeys?.length > 0 ? apiKeys[0].key : null) ||
-        (!cloudEnabled ? "sk_omniroute" : null);
+      // (#523) Prefer keyId lookup so the backend writes the real key to disk.
+      // Fall back to sk_omniroute for localhost-only setups without a key.
+      const selectedKeyId = selectedApiKey?.trim() || (apiKeys?.length > 0 ? apiKeys[0].id : null);
+      const skOmnirouteFallback = !cloudEnabled ? "sk_omniroute" : null;
 
-      if (keyToUse) {
-        env.ANTHROPIC_AUTH_TOKEN = keyToUse;
+      if (!selectedKeyId && skOmnirouteFallback) {
+        env.ANTHROPIC_AUTH_TOKEN = skOmnirouteFallback;
       }
 
       tool.defaultModels.forEach((model) => {
         const targetModel = modelMappings[model.alias];
         if (targetModel && model.envKey) env[model.envKey] = targetModel;
       });
+
+      const postBody: Record<string, unknown> = { env };
+      if (selectedKeyId) postBody.keyId = selectedKeyId;
+
       const res = await fetch("/api/cli-tools/claude-settings", {
         method: "POST",
         headers: { "Content-Type": "application/json" },
-        body: JSON.stringify({ env }),
+        body: JSON.stringify(postBody),
       });
       const data = await res.json();
       if (res.ok) {
@@ -412,7 +418,7 @@ export default function ClaudeToolCard({
                       className="flex-1 px-2 py-1.5 bg-surface rounded text-xs border border-border focus:outline-none focus:ring-1 focus:ring-primary/50"
                     >
                       {apiKeys.map((key) => (
-                        <option key={key.id} value={key.key}>
+                        <option key={key.id} value={key.id}>
                           {key.key}
                         </option>
                       ))}

src/app/(dashboard)/dashboard/providers/[id]/page.tsx

@@ -1,6 +1,6 @@
 "use client";
 
-import { useState, useEffect, useCallback, useRef } from "react";
+import { useState, useEffect, useCallback, useRef, useMemo } from "react";
 import { useNotificationStore } from "@/store/notificationStore";
 import PropTypes from "prop-types";
 import { useParams, useRouter } from "next/navigation";
@@ -31,6 +31,129 @@ import {
 } from "@/shared/constants/providers";
 import { getModelsByProviderId } from "@/shared/constants/models";
 import { useCopyToClipboard } from "@/shared/hooks/useCopyToClipboard";
+import {
+  MODEL_COMPAT_PROTOCOL_KEYS,
+  type ModelCompatProtocolKey,
+} from "@/shared/constants/modelCompat";
+
+type CompatByProtocolMap = Partial<
+  Record<
+    ModelCompatProtocolKey,
+    { normalizeToolCallId?: boolean; preserveOpenAIDeveloperRole?: boolean }
+  >
+>;
+type CompatModelRow = {
+  id?: string;
+  name?: string;
+  source?: string;
+  apiFormat?: string;
+  supportedEndpoints?: string[];
+  normalizeToolCallId?: boolean;
+  preserveOpenAIDeveloperRole?: boolean;
+  compatByProtocol?: CompatByProtocolMap;
+};
+
+type CompatModelMap = Map<string, CompatModelRow>;
+
+function buildCompatMap(rows: CompatModelRow[]): CompatModelMap {
+  const m = new Map<string, CompatModelRow>();
+  for (const r of rows) if (r.id) m.set(r.id, r);
+  return m;
+}
+
+function getProtoSlice(
+  c: CompatModelRow | undefined,
+  o: CompatModelRow | undefined,
+  protocol: string
+) {
+  return c?.compatByProtocol?.[protocol] ?? o?.compatByProtocol?.[protocol];
+}
+
+function effectiveNormalizeForProtocol(
+  modelId: string,
+  protocol: string,
+  customMap: CompatModelMap,
+  overrideMap: CompatModelMap
+): boolean {
+  const c = customMap.get(modelId);
+  const o = overrideMap.get(modelId);
+  const pc = getProtoSlice(c, o, protocol);
+  if (pc && Object.prototype.hasOwnProperty.call(pc, "normalizeToolCallId")) {
+    return Boolean(pc.normalizeToolCallId);
+  }
+  if (c?.normalizeToolCallId) return true;
+  return Boolean(o?.normalizeToolCallId);
+}
+
+function effectivePreserveForProtocol(
+  modelId: string,
+  protocol: string,
+  customMap: CompatModelMap,
+  overrideMap: CompatModelMap
+): boolean {
+  const c = customMap.get(modelId);
+  const o = overrideMap.get(modelId);
+  const pc = getProtoSlice(c, o, protocol);
+  if (pc && Object.prototype.hasOwnProperty.call(pc, "preserveOpenAIDeveloperRole")) {
+    return Boolean(pc.preserveOpenAIDeveloperRole);
+  }
+  if (c && Object.prototype.hasOwnProperty.call(c, "preserveOpenAIDeveloperRole")) {
+    return Boolean(c.preserveOpenAIDeveloperRole);
+  }
+  if (o && Object.prototype.hasOwnProperty.call(o, "preserveOpenAIDeveloperRole")) {
+    return Boolean(o.preserveOpenAIDeveloperRole);
+  }
+  return true;
+}
+
+function anyNormalizeCompatBadge(
+  modelId: string,
+  customMap: CompatModelMap,
+  overrideMap: CompatModelMap
+): boolean {
+  const c = customMap.get(modelId);
+  const o = overrideMap.get(modelId);
+  if (c?.normalizeToolCallId || o?.normalizeToolCallId) return true;
+  for (const p of MODEL_COMPAT_PROTOCOL_KEYS) {
+    const pc = getProtoSlice(c, o, p);
+    if (pc?.normalizeToolCallId) return true;
+  }
+  return false;
+}
+
+function anyNoPreserveCompatBadge(
+  modelId: string,
+  customMap: CompatModelMap,
+  overrideMap: CompatModelMap
+): boolean {
+  const c = customMap.get(modelId);
+  const o = overrideMap.get(modelId);
+  if (
+    c &&
+    Object.prototype.hasOwnProperty.call(c, "preserveOpenAIDeveloperRole") &&
+    c.preserveOpenAIDeveloperRole === false
+  ) {
+    return true;
+  }
+  if (
+    o &&
+    Object.prototype.hasOwnProperty.call(o, "preserveOpenAIDeveloperRole") &&
+    o.preserveOpenAIDeveloperRole === false
+  ) {
+    return true;
+  }
+  for (const p of MODEL_COMPAT_PROTOCOL_KEYS) {
+    const pc = getProtoSlice(c, o, p);
+    if (
+      pc &&
+      Object.prototype.hasOwnProperty.call(pc, "preserveOpenAIDeveloperRole") &&
+      pc.preserveOpenAIDeveloperRole === false
+    ) {
+      return true;
+    }
+  }
+  return false;
+}
 
 interface ModelRowProps {
   model: { id: string };
@@ -40,10 +163,16 @@ interface ModelRowProps {
   onCopy: (text: string, key: string) => void;
   t: (key: string, values?: Record<string, unknown>) => string;
   showDeveloperToggle?: boolean;
-  normalizeToolCallId?: boolean;
-  preserveDeveloperRole?: boolean;
-  onNormalizeChange?: (v: boolean) => void;
-  onPreserveChange?: (v: boolean) => void;
+  effectiveModelNormalize: (modelId: string, protocol?: string) => boolean;
+  effectiveModelPreserveDeveloper: (modelId: string, protocol?: string) => boolean;
+  saveModelCompatFlags: (
+    modelId: string,
+    patch: {
+      normalizeToolCallId?: boolean;
+      preserveOpenAIDeveloperRole?: boolean;
+      compatByProtocol?: CompatByProtocolMap;
+    }
+  ) => void;
   compatDisabled?: boolean;
 }
 
@@ -55,10 +184,16 @@ interface PassthroughModelRowProps {
   onDeleteAlias: () => void;
   t: (key: string, values?: Record<string, unknown>) => string;
   showDeveloperToggle?: boolean;
-  normalizeToolCallId?: boolean;
-  preserveDeveloperRole?: boolean;
-  onNormalizeChange?: (v: boolean) => void;
-  onPreserveChange?: (v: boolean) => void;
+  effectiveModelNormalize: (modelId: string, protocol?: string) => boolean;
+  effectiveModelPreserveDeveloper: (modelId: string, protocol?: string) => boolean;
+  saveModelCompatFlags: (
+    modelId: string,
+    patch: {
+      normalizeToolCallId?: boolean;
+      preserveOpenAIDeveloperRole?: boolean;
+      compatByProtocol?: CompatByProtocolMap;
+    }
+  ) => void;
   compatDisabled?: boolean;
 }
 
@@ -230,27 +365,43 @@ function normalizeCodexLimitPolicy(policy: unknown): { use5h: boolean; useWeekly
   };
 }
 
+function compatProtocolLabelKey(protocol: string): string {
+  if (protocol === "openai") return "compatProtocolOpenAI";
+  if (protocol === "openai-responses") return "compatProtocolOpenAIResponses";
+  if (protocol === "claude") return "compatProtocolClaude";
+  return "compatProtocolOpenAI";
+}
+
 function ModelCompatPopover({
   t,
-  normalizeToolCallId,
-  preserveDeveloperRole,
+  effectiveModelNormalize,
+  effectiveModelPreserveDeveloper,
+  onCompatPatch,
   showDeveloperToggle = true,
-  onNormalizeChange,
-  onPreserveChange,
   disabled,
 }: {
   t: (key: string) => string;
-  normalizeToolCallId: boolean;
-  preserveDeveloperRole?: boolean;
+  effectiveModelNormalize: (protocol: string) => boolean;
+  effectiveModelPreserveDeveloper: (protocol: string) => boolean;
+  onCompatPatch: (
+    protocol: string,
+    payload: {
+      normalizeToolCallId?: boolean;
+      preserveOpenAIDeveloperRole?: boolean;
+    }
+  ) => void;
   showDeveloperToggle?: boolean;
-  onNormalizeChange: (v: boolean) => void;
-  onPreserveChange: (v: boolean) => void;
   disabled?: boolean;
 }) {
   const [open, setOpen] = useState(false);
+  const [protocol, setProtocol] = useState<string>(MODEL_COMPAT_PROTOCOL_KEYS[0]);
   const ref = useRef<HTMLDivElement>(null);
   const panelRef = useRef<HTMLDivElement | null>(null);
 
+  const normalizeToolCallId = effectiveModelNormalize(protocol);
+  const preserveDeveloperRole = effectiveModelPreserveDeveloper(protocol);
+  const devToggle = showDeveloperToggle && protocol !== "claude";
+
   // Click-outside: check both trigger and panel so that if the panel is ever rendered
   // in a portal (outside this subtree), clicks inside the panel still do not close it.
   useEffect(() => {
@@ -280,32 +431,47 @@ function ModelCompatPopover({
       {open && (
         <div
           ref={panelRef}
-          className="absolute left-0 top-full mt-1 z-50 min-w-[200px] p-3 rounded-lg border border-border bg-white dark:bg-zinc-900 shadow-xl ring-1 ring-black/5 dark:ring-white/10"
+          className="absolute left-0 top-full mt-1 z-50 min-w-[220px] max-w-[92vw] p-3 rounded-lg border border-border bg-white dark:bg-zinc-900 shadow-xl ring-1 ring-black/5 dark:ring-white/10"
         >
-          <p className="text-[10px] font-semibold uppercase tracking-wide text-text-muted mb-2">
+          <p className="text-[10px] font-semibold uppercase tracking-wide text-text-muted mb-1">
             {t("compatAdjustmentsTitle")}
           </p>
+          <p className="text-[10px] text-text-muted mb-2 leading-snug">{t("compatProtocolHint")}</p>
+          <label className="block text-[10px] font-medium text-text-muted mb-1">
+            {t("compatProtocolLabel")}
+          </label>
+          <select
+            value={protocol}
+            onChange={(e) => setProtocol(e.target.value)}
+            disabled={disabled}
+            className="w-full mb-3 px-2 py-1.5 text-xs rounded-md border border-border bg-white dark:bg-zinc-800 text-zinc-900 dark:text-zinc-100 focus:outline-none focus:ring-1 focus:ring-primary/50"
+          >
+            {MODEL_COMPAT_PROTOCOL_KEYS.map((p) => (
+              <option key={p} value={p}>
+                {t(compatProtocolLabelKey(p))}
+              </option>
+            ))}
+          </select>
           <div className="flex flex-col gap-3">
             <Toggle
               size="sm"
               label={t("compatToolIdShort")}
               title={t("normalizeToolCallIdLabel")}
               checked={normalizeToolCallId}
-              onChange={onNormalizeChange}
+              onChange={(v) => onCompatPatch(protocol, { normalizeToolCallId: v })}
               disabled={disabled}
             />
-            {showDeveloperToggle && (
-              <>
-                {/* Inversion: Toggle checked = "do not preserve" (UI). onPreserveChange(val) = value to store (true = preserve, false = do not preserve), so we pass !checked. */}
-                <Toggle
-                  size="sm"
-                  label={t("compatDoNotPreserveDeveloper")}
-                  title={t("preserveDeveloperRoleLabel")}
-                  checked={preserveDeveloperRole === false}
-                  onChange={(checked) => onPreserveChange(!checked)}
-                  disabled={disabled}
-                />
-              </>
+            {devToggle && (
+              <Toggle
+                size="sm"
+                label={t("compatDoNotPreserveDeveloper")}
+                title={t("preserveDeveloperRoleLabel")}
+                checked={preserveDeveloperRole === false}
+                onChange={(checked) =>
+                  onCompatPatch(protocol, { preserveOpenAIDeveloperRole: !checked })
+                }
+                disabled={disabled}
+              />
             )}
           </div>
         </div>
@@ -350,12 +516,8 @@ export default function ProviderDetailPage() {
     importedCount: 0,
   });
   const [modelMeta, setModelMeta] = useState<{
-    customModels: Record<string, unknown>[];
-    modelCompatOverrides: {
-      id: string;
-      normalizeToolCallId?: boolean;
-      preserveOpenAIDeveloperRole?: boolean;
-    }[];
+    customModels: CompatModelRow[];
+    modelCompatOverrides: Array<CompatModelRow & { id: string }>;
   }>({ customModels: [], modelCompatOverrides: [] });
   const [compatSavingModelId, setCompatSavingModelId] = useState<string | null>(null);
 
@@ -1020,48 +1182,91 @@ export default function ProviderDetailPage() {
 
   const canImportModels = connections.some((conn) => conn.isActive !== false);
 
-  const effectiveModelNormalize = (modelId: string) => {
-    const c = modelMeta.customModels.find((m: { id?: string }) => m.id === modelId) as
-      | { normalizeToolCallId?: boolean }
-      | undefined;
-    if (c) return Boolean(c.normalizeToolCallId);
-    const o = modelMeta.modelCompatOverrides.find((e) => e.id === modelId);
-    return Boolean(o?.normalizeToolCallId);
-  };
+  const customMap = useMemo(() => buildCompatMap(modelMeta.customModels), [modelMeta.customModels]);
+  const overrideMap = useMemo(
+    () => buildCompatMap(modelMeta.modelCompatOverrides),
+    [modelMeta.modelCompatOverrides]
+  );
 
-  const effectiveModelPreserveDeveloper = (modelId: string) => {
-    const c = modelMeta.customModels.find((m: { id?: string }) => m.id === modelId) as
-      | Record<string, unknown>
-      | undefined;
-    if (c && Object.prototype.hasOwnProperty.call(c, "preserveOpenAIDeveloperRole")) {
-      return Boolean(c.preserveOpenAIDeveloperRole);
-    }
-    const o = modelMeta.modelCompatOverrides.find((e) => e.id === modelId);
-    if (o && Object.prototype.hasOwnProperty.call(o, "preserveOpenAIDeveloperRole")) {
-      return Boolean(o.preserveOpenAIDeveloperRole);
-    }
-    return true;
-  };
+  const effectiveModelNormalize = (modelId: string, protocol = MODEL_COMPAT_PROTOCOL_KEYS[0]) =>
+    effectiveNormalizeForProtocol(modelId, protocol, customMap, overrideMap);
+
+  const effectiveModelPreserveDeveloper = (
+    modelId: string,
+    protocol = MODEL_COMPAT_PROTOCOL_KEYS[0]
+  ) => effectivePreserveForProtocol(modelId, protocol, customMap, overrideMap);
 
   const saveModelCompatFlags = async (
     modelId: string,
-    patch: { normalizeToolCallId?: boolean; preserveOpenAIDeveloperRole?: boolean }
+    patch: {
+      normalizeToolCallId?: boolean;
+      preserveOpenAIDeveloperRole?: boolean;
+      compatByProtocol?: CompatByProtocolMap;
+    }
   ) => {
     setCompatSavingModelId(modelId);
     try {
-      const body: Record<string, unknown> = { provider: providerId, modelId, ...patch };
+      const c = customMap.get(modelId) as Record<string, unknown> | undefined;
+      let body: Record<string, unknown>;
+      const onlyCompatByProtocol =
+        patch.compatByProtocol &&
+        patch.normalizeToolCallId === undefined &&
+        patch.preserveOpenAIDeveloperRole === undefined;
+
+      if (c) {
+        if (onlyCompatByProtocol) {
+          body = {
+            provider: providerId,
+            modelId,
+            compatByProtocol: patch.compatByProtocol,
+          };
+        } else {
+          body = {
+            provider: providerId,
+            modelId,
+            modelName: (c.name as string) || modelId,
+            source: (c.source as string) || "manual",
+            apiFormat: (c.apiFormat as string) || "chat-completions",
+            supportedEndpoints:
+              Array.isArray(c.supportedEndpoints) && (c.supportedEndpoints as unknown[]).length
+                ? c.supportedEndpoints
+                : ["chat"],
+            normalizeToolCallId:
+              patch.normalizeToolCallId !== undefined
+                ? patch.normalizeToolCallId
+                : Boolean(c.normalizeToolCallId),
+            preserveOpenAIDeveloperRole:
+              patch.preserveOpenAIDeveloperRole !== undefined
+                ? patch.preserveOpenAIDeveloperRole
+                : Object.prototype.hasOwnProperty.call(c, "preserveOpenAIDeveloperRole")
+                  ? Boolean(c.preserveOpenAIDeveloperRole)
+                  : true,
+          };
+          if (patch.compatByProtocol) body.compatByProtocol = patch.compatByProtocol;
+        }
+      } else {
+        body = { provider: providerId, modelId, ...patch };
+      }
       const res = await fetch("/api/provider-models", {
         method: "PUT",
         headers: { "Content-Type": "application/json" },
         body: JSON.stringify(body),
       });
-      if (res.ok) await fetchProviderModelMeta();
-      else notify.error(t("failedSaveCustomModel"));
+      if (!res.ok) {
+        notify.error(t("failedSaveCustomModel"));
+        return;
+      }
     } catch {
       notify.error(t("failedSaveCustomModel"));
+      return;
     } finally {
       setCompatSavingModelId(null);
     }
+    try {
+      await fetchProviderModelMeta();
+    } catch {
+      /* refresh failure is non-critical — data was already saved */
+    }
   };
 
   const renderModelsSection = () => {
@@ -1166,14 +1371,9 @@ export default function ProviderDetailPage() {
                 onCopy={copy}
                 t={t}
                 showDeveloperToggle
-                normalizeToolCallId={effectiveModelNormalize(model.id)}
-                preserveDeveloperRole={effectiveModelPreserveDeveloper(model.id)}
-                onNormalizeChange={(v) =>
-                  saveModelCompatFlags(model.id, { normalizeToolCallId: v })
-                }
-                onPreserveChange={(v) =>
-                  saveModelCompatFlags(model.id, { preserveOpenAIDeveloperRole: v })
-                }
+                effectiveModelNormalize={effectiveModelNormalize}
+                effectiveModelPreserveDeveloper={effectiveModelPreserveDeveloper}
+                saveModelCompatFlags={saveModelCompatFlags}
                 compatDisabled={compatSavingModelId === model.id}
               />
             );
@@ -1813,10 +2013,9 @@ function ModelRow({
   onCopy,
   t,
   showDeveloperToggle = true,
-  normalizeToolCallId,
-  preserveDeveloperRole,
-  onNormalizeChange,
-  onPreserveChange,
+  effectiveModelNormalize,
+  effectiveModelPreserveDeveloper,
+  saveModelCompatFlags,
   compatDisabled,
 }: ModelRowProps) {
   return (
@@ -1840,11 +2039,12 @@ function ModelRow({
       </div>
       <ModelCompatPopover
         t={t}
-        normalizeToolCallId={Boolean(normalizeToolCallId)}
-        preserveDeveloperRole={preserveDeveloperRole}
+        effectiveModelNormalize={(p) => effectiveModelNormalize(model.id, p)}
+        effectiveModelPreserveDeveloper={(p) => effectiveModelPreserveDeveloper(model.id, p)}
+        onCompatPatch={(protocol, payload) =>
+          saveModelCompatFlags(model.id, { compatByProtocol: { [protocol]: payload } })
+        }
         showDeveloperToggle={showDeveloperToggle}
-        onNormalizeChange={onNormalizeChange}
-        onPreserveChange={onPreserveChange}
         disabled={compatDisabled}
       />
     </div>
@@ -1861,10 +2061,9 @@ ModelRow.propTypes = {
   onCopy: PropTypes.func.isRequired,
   t: PropTypes.func,
   showDeveloperToggle: PropTypes.bool,
-  normalizeToolCallId: PropTypes.bool,
-  preserveDeveloperRole: PropTypes.bool,
-  onNormalizeChange: PropTypes.func,
-  onPreserveChange: PropTypes.func,
+  effectiveModelNormalize: PropTypes.func.isRequired,
+  effectiveModelPreserveDeveloper: PropTypes.func.isRequired,
+  saveModelCompatFlags: PropTypes.func.isRequired,
   compatDisabled: PropTypes.bool,
 };
 
@@ -1960,12 +2159,9 @@ function PassthroughModelsSection({
               onDeleteAlias={() => onDeleteAlias(alias)}
               t={t}
               showDeveloperToggle
-              normalizeToolCallId={effectiveModelNormalize(modelId)}
-              preserveDeveloperRole={effectiveModelPreserveDeveloper(modelId)}
-              onNormalizeChange={(v) => saveModelCompatFlags(modelId, { normalizeToolCallId: v })}
-              onPreserveChange={(v) =>
-                saveModelCompatFlags(modelId, { preserveOpenAIDeveloperRole: v })
-              }
+              effectiveModelNormalize={effectiveModelNormalize}
+              effectiveModelPreserveDeveloper={effectiveModelPreserveDeveloper}
+              saveModelCompatFlags={saveModelCompatFlags}
               compatDisabled={compatSavingModelId === modelId}
             />
           ))}
@@ -1997,10 +2193,9 @@ function PassthroughModelRow({
   onDeleteAlias,
   t,
   showDeveloperToggle = true,
-  normalizeToolCallId,
-  preserveDeveloperRole,
-  onNormalizeChange,
-  onPreserveChange,
+  effectiveModelNormalize,
+  effectiveModelPreserveDeveloper,
+  saveModelCompatFlags,
   compatDisabled,
 }: PassthroughModelRowProps) {
   return (
@@ -2037,11 +2232,12 @@ function PassthroughModelRow({
       <div className="pl-9">
         <ModelCompatPopover
           t={t}
-          normalizeToolCallId={Boolean(normalizeToolCallId)}
-          preserveDeveloperRole={preserveDeveloperRole}
+          effectiveModelNormalize={(p) => effectiveModelNormalize(modelId, p)}
+          effectiveModelPreserveDeveloper={(p) => effectiveModelPreserveDeveloper(modelId, p)}
+          onCompatPatch={(protocol, payload) =>
+            saveModelCompatFlags(modelId, { compatByProtocol: { [protocol]: payload } })
+          }
           showDeveloperToggle={showDeveloperToggle}
-          onNormalizeChange={onNormalizeChange}
-          onPreserveChange={onPreserveChange}
           disabled={compatDisabled}
         />
       </div>
@@ -2057,10 +2253,9 @@ PassthroughModelRow.propTypes = {
   onDeleteAlias: PropTypes.func.isRequired,
   t: PropTypes.func,
   showDeveloperToggle: PropTypes.bool,
-  normalizeToolCallId: PropTypes.bool,
-  preserveDeveloperRole: PropTypes.bool,
-  onNormalizeChange: PropTypes.func,
-  onPreserveChange: PropTypes.func,
+  effectiveModelNormalize: PropTypes.func.isRequired,
+  effectiveModelPreserveDeveloper: PropTypes.func.isRequired,
+  saveModelCompatFlags: PropTypes.func.isRequired,
   compatDisabled: PropTypes.bool,
 };
 
@@ -2075,7 +2270,10 @@ function CustomModelsSection({
 }: CustomModelsSectionProps) {
   const t = useTranslations("providers");
   const notify = useNotificationStore();
-  const [customModels, setCustomModels] = useState([]);
+  const [customModels, setCustomModels] = useState<CompatModelRow[]>([]);
+  const [modelCompatOverrides, setModelCompatOverrides] = useState<
+    Array<CompatModelRow & { id: string }>
+  >([]);
   const [newModelId, setNewModelId] = useState("");
   const [newModelName, setNewModelName] = useState("");
   const [newApiFormat, setNewApiFormat] = useState("chat-completions");
@@ -2085,16 +2283,18 @@ function CustomModelsSection({
   const [editingModelId, setEditingModelId] = useState<string | null>(null);
   const [editingApiFormat, setEditingApiFormat] = useState("chat-completions");
   const [editingEndpoints, setEditingEndpoints] = useState<string[]>(["chat"]);
-  const [editingNormalizeToolCallId, setEditingNormalizeToolCallId] = useState(false);
-  const [editingPreserveDeveloperRole, setEditingPreserveDeveloperRole] = useState(false);
   const [savingModelId, setSavingModelId] = useState<string | null>(null);
 
+  const customMap = useMemo(() => buildCompatMap(customModels), [customModels]);
+  const overrideMap = useMemo(() => buildCompatMap(modelCompatOverrides), [modelCompatOverrides]);
+
   const fetchCustomModels = useCallback(async () => {
     try {
       const res = await fetch(`/api/provider-models?provider=${encodeURIComponent(providerId)}`);
       if (res.ok) {
         const data = await res.json();
         setCustomModels(data.models || []);
+        setModelCompatOverrides(data.modelCompatOverrides || []);
       }
     } catch (e) {
       console.error("Failed to fetch custom models:", e);
@@ -2160,23 +2360,44 @@ function CustomModelsSection({
         ? model.supportedEndpoints
         : ["chat"]
     );
-    setEditingNormalizeToolCallId(Boolean(model.normalizeToolCallId));
-    setEditingPreserveDeveloperRole(
-      Object.prototype.hasOwnProperty.call(model, "preserveOpenAIDeveloperRole")
-        ? Boolean(model.preserveOpenAIDeveloperRole)
-        : true
-    );
   };
 
   const cancelEdit = () => {
     setEditingModelId(null);
     setEditingApiFormat("chat-completions");
     setEditingEndpoints(["chat"]);
-    setEditingNormalizeToolCallId(false);
-    setEditingPreserveDeveloperRole(true);
     setSavingModelId(null);
   };
 
+  const saveCustomCompat = async (
+    modelId: string,
+    patch: { compatByProtocol?: CompatByProtocolMap }
+  ) => {
+    setSavingModelId(modelId);
+    try {
+      const res = await fetch("/api/provider-models", {
+        method: "PUT",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify({ provider: providerId, modelId, ...patch }),
+      });
+      if (!res.ok) {
+        notify.error(t("failedSaveCustomModel"));
+        return;
+      }
+    } catch {
+      notify.error(t("failedSaveCustomModel"));
+      return;
+    } finally {
+      setSavingModelId(null);
+    }
+    try {
+      await fetchCustomModels();
+      onModelsChanged?.();
+    } catch {
+      /* refresh failure is non-critical — data was already saved */
+    }
+  };
+
   const saveEdit = async (modelId) => {
     if (!editingModelId || editingModelId !== modelId) return;
     if (!editingEndpoints.length) {
@@ -2197,8 +2418,6 @@ function CustomModelsSection({
           source: model?.source || "manual",
           apiFormat: editingApiFormat,
           supportedEndpoints: editingEndpoints,
-          normalizeToolCallId: editingNormalizeToolCallId,
-          preserveOpenAIDeveloperRole: editingPreserveDeveloperRole,
         }),
       });
 
@@ -2361,7 +2580,7 @@ function CustomModelsSection({
                         🔊 Audio
                       </span>
                     )}
-                    {model.normalizeToolCallId && (
+                    {anyNormalizeCompatBadge(model.id, customMap, overrideMap) && (
                       <span
                         className="text-[10px] px-1.5 py-0.5 rounded-full bg-slate-500/15 text-slate-400 font-medium"
                         title={t("normalizeToolCallIdLabel")}
@@ -2369,7 +2588,7 @@ function CustomModelsSection({
                         ID×9
                       </span>
                     )}
-                    {model.preserveOpenAIDeveloperRole === false && (
+                    {anyNoPreserveCompatBadge(model.id, customMap, overrideMap) && (
                       <span
                         className="text-[10px] px-1.5 py-0.5 rounded-full bg-cyan-500/15 text-cyan-400 font-medium"
                         title={t("compatDoNotPreserveDeveloper")}
@@ -2433,11 +2652,18 @@ function CustomModelsSection({
                       <div className="mt-3 pt-3 border-t border-border/80 w-full">
                         <ModelCompatPopover
                           t={t}
-                          normalizeToolCallId={editingNormalizeToolCallId}
-                          preserveDeveloperRole={editingPreserveDeveloperRole}
+                          effectiveModelNormalize={(p) =>
+                            effectiveNormalizeForProtocol(model.id, p, customMap, overrideMap)
+                          }
+                          effectiveModelPreserveDeveloper={(p) =>
+                            effectivePreserveForProtocol(model.id, p, customMap, overrideMap)
+                          }
+                          onCompatPatch={(protocol, payload) =>
+                            saveCustomCompat(model.id, {
+                              compatByProtocol: { [protocol]: payload },
+                            })
+                          }
                           showDeveloperToggle
-                          onNormalizeChange={setEditingNormalizeToolCallId}
-                          onPreserveChange={setEditingPreserveDeveloperRole}
                           disabled={savingModelId === model.id}
                         />
                       </div>
@@ -2716,12 +2942,9 @@ function CompatibleModelsSection({
               onDeleteAlias={() => handleDeleteModel(modelId, alias)}
               t={t}
               showDeveloperToggle={!isAnthropic}
-              normalizeToolCallId={effectiveModelNormalize(modelId)}
-              preserveDeveloperRole={effectiveModelPreserveDeveloper(modelId)}
-              onNormalizeChange={(v) => saveModelCompatFlags(modelId, { normalizeToolCallId: v })}
-              onPreserveChange={(v) =>
-                saveModelCompatFlags(modelId, { preserveOpenAIDeveloperRole: v })
-              }
+              effectiveModelNormalize={effectiveModelNormalize}
+              effectiveModelPreserveDeveloper={effectiveModelPreserveDeveloper}
+              saveModelCompatFlags={saveModelCompatFlags}
               compatDisabled={compatSavingModelId === modelId}
             />
           ))}

src/app/(dashboard)/dashboard/providers/page.tsx

@@ -2,6 +2,7 @@
 
 import { useState, useEffect } from "react";
 import Image from "next/image";
+import ProviderIcon from "@/shared/components/ProviderIcon";
 import PropTypes from "prop-types";
 import {
   Card,
@@ -490,16 +491,8 @@ function ProviderCard({ providerId, provider, stats, authType, onToggle }) {
   const t = useTranslations("providers");
   const tc = useTranslations("common");
   const { connected, error, errorCode, errorTime, allDisabled } = stats;
-  const [imgSrc, setImgSrc] = useState(`/providers/${provider.id}.png`);
-  const [imgError, setImgError] = useState(false);
 
-  const handleImgError = () => {
-    if (imgSrc.endsWith(".png")) {
-      setImgSrc(`/providers/${provider.id}.svg`);
-    } else {
-      setImgError(true);
-    }
-  };
+  // (#529) Icon state replaced by ProviderIcon component (Lobehub + PNG + generic fallback)
 
   const dotColors = {
     free: "bg-green-500",
@@ -526,21 +519,8 @@ function ProviderCard({ providerId, provider, stats, authType, onToggle }) {
               className="size-8 rounded-lg flex items-center justify-center"
               style={{ backgroundColor: `${provider.color}15` }}
             >
-              {imgError ? (
-                <span className="text-xs font-bold" style={{ color: provider.color }}>
-                  {provider.textIcon || provider.id.slice(0, 2).toUpperCase()}
-                </span>
-              ) : (
-                <Image
-                  src={imgSrc}
-                  alt={provider.name}
-                  width={30}
-                  height={30}
-                  className="object-contain rounded-lg max-w-[32px] max-h-[32px]"
-                  sizes="32px"
-                  onError={handleImgError}
-                />
-              )}
+              {/* (#529) ProviderIcon: Lobehub icons → PNG fallback → generic icon */}
+              <ProviderIcon providerId={provider.id} size={28} type="color" />
             </div>
             <div>
               <h3 className="font-semibold flex items-center gap-1.5">
@@ -633,28 +613,15 @@ function ApiKeyProviderCard({ providerId, provider, stats, authType, onToggle })
     compatible: t("compatibleLabel"),
   };
 
-  // Determine icon path: OpenAI Compatible providers use specialized icons
-  const getIconPath = () => {
+  // (#529) Icon state replaced by ProviderIcon component
+  // For compatible/anthropic providers, continue using static PNGs via the icon path
+  const staticIconPath = (() => {
     if (isCompatible) {
       return provider.apiType === "responses" ? "/providers/oai-r.png" : "/providers/oai-cc.png";
     }
-    if (isAnthropicCompatible) {
-      return "/providers/anthropic-m.png"; // Use Anthropic icon as base
-    }
-    return `/providers/${provider.id}.png`;
-  };
-
-  const [imgSrc, setImgSrc] = useState<string>(() => getIconPath());
-  const [imgError, setImgError] = useState(false);
-
-  const handleImgError = () => {
-    const basePath = getIconPath();
-    if (imgSrc.endsWith(".png") && !isCompatible && !isAnthropicCompatible) {
-      setImgSrc(`/providers/${provider.id}.svg`);
-    } else {
-      setImgError(true);
-    }
-  };
+    if (isAnthropicCompatible) return "/providers/anthropic-m.png";
+    return null; // ProviderIcon will handle it
+  })();
 
   return (
     <Link href={`/dashboard/providers/${providerId}`} className="group">
@@ -668,20 +635,18 @@ function ApiKeyProviderCard({ providerId, provider, stats, authType, onToggle })
               className="size-8 rounded-lg flex items-center justify-center"
               style={{ backgroundColor: `${provider.color}15` }}
             >
-              {imgError ? (
-                <span className="text-xs font-bold" style={{ color: provider.color }}>
-                  {provider.textIcon || provider.id.slice(0, 2).toUpperCase()}
-                </span>
-              ) : (
+              {/* (#529) ProviderIcon with static override for compatible providers */}
+              {staticIconPath ? (
                 <Image
-                  src={imgSrc || getIconPath()}
+                  src={staticIconPath}
                   alt={provider.name}
                   width={30}
                   height={30}
                   className="object-contain rounded-lg max-w-[30px] max-h-[30px]"
                   sizes="30px"
-                  onError={handleImgError}
                 />
+              ) : (
+                <ProviderIcon providerId={provider.id} size={28} type="color" />
               )}
             </div>
             <div>

src/app/api/cli-tools/claude-settings/route.ts

@@ -12,6 +12,7 @@ import { createBackup } from "@/shared/services/backupService";
 import { saveCliToolLastConfigured, deleteCliToolLastConfigured } from "@/lib/db/cliToolState";
 import { cliSettingsEnvSchema } from "@/shared/validation/schemas";
 import { isValidationFailure, validateBody } from "@/shared/validation/helpers";
+import { getApiKeyById } from "@/lib/localDb";
 
 // Get claude settings path based on OS
 const getClaudeSettingsPath = () => getCliPrimaryConfigPath("claude");
@@ -100,6 +101,22 @@ export async function POST(request: Request) {
     }
     const { env } = validation.data;
 
+    // (#523/#526) If a keyId was provided, resolve the real API key from DB.
+    // The /api/keys list endpoint returns masked key strings — sending those to
+    // disk would save an unusable half-hidden token. Resolving by ID guarantees
+    // we always write the full key value to the config file.
+    const keyId = typeof rawBody?.keyId === "string" ? rawBody.keyId.trim() : null;
+    if (keyId) {
+      try {
+        const keyRecord = await getApiKeyById(keyId);
+        if (keyRecord?.key) {
+          env.ANTHROPIC_AUTH_TOKEN = keyRecord.key as string;
+        }
+      } catch {
+        // Non-critical: fall back to whatever value was in env (e.g. sk_omniroute)
+      }
+    }
+
     const settingsPath = getClaudeSettingsPath();
     const claudeDir = path.dirname(settingsPath);
 

src/app/api/cli-tools/cline-settings/route.ts

@@ -9,6 +9,7 @@ import { createBackup } from "@/shared/services/backupService";
 import { saveCliToolLastConfigured, deleteCliToolLastConfigured } from "@/lib/db/cliToolState";
 import { cliModelConfigSchema } from "@/shared/validation/schemas";
 import { isValidationFailure, validateBody } from "@/shared/validation/helpers";
+import { getApiKeyById } from "@/lib/localDb";
 
 const CLINE_DATA_DIR = path.join(os.homedir(), ".cline", "data");
 const GLOBAL_STATE_PATH = path.join(CLINE_DATA_DIR, "globalState.json");
@@ -125,7 +126,18 @@ export async function POST(request: Request) {
     if (isValidationFailure(validation)) {
       return NextResponse.json({ error: validation.error }, { status: 400 });
     }
-    const { baseUrl, apiKey, model } = validation.data;
+    let { baseUrl, apiKey, model } = validation.data;
+
+    // (#526) Resolve real key from DB if keyId was provided
+    const keyId = typeof rawBody?.keyId === "string" ? rawBody.keyId.trim() : null;
+    if (keyId) {
+      try {
+        const keyRecord = await getApiKeyById(keyId);
+        if (keyRecord?.key) apiKey = keyRecord.key as string;
+      } catch {
+        /* non-critical */
+      }
+    }
 
     // Ensure directory exists
     await fs.mkdir(CLINE_DATA_DIR, { recursive: true });

src/app/api/cli-tools/guide-settings/[toolId]/route.ts

@@ -39,6 +39,10 @@ export async function POST(request, { params }) {
     switch (toolId) {
       case "continue":
         return await saveContinueConfig({ baseUrl, apiKey, model });
+      case "opencode":
+        // (#524) OpenCode config was never saved because only 'continue' was handled here.
+        // opencode reads ~/.config/opencode/config.toml — write the OmniRoute settings there.
+        return await saveOpenCodeConfig({ baseUrl, apiKey, model });
       default:
         return NextResponse.json(
           { error: `Direct config save not supported for: ${toolId}` },
@@ -125,3 +129,56 @@ async function saveContinueConfig({ baseUrl, apiKey, model }) {
     configPath,
   });
 }
+
+/**
+ * Save OpenCode config to ~/.config/opencode/config.toml (XDG_CONFIG_HOME aware).
+ * (#524) OpenCode was silently failing because this handler was missing.
+ */
+async function saveOpenCodeConfig({ baseUrl, apiKey, model }) {
+  const { apiPort } = getRuntimePorts();
+  // Honour $XDG_CONFIG_HOME if set, otherwise use ~/.config per the XDG Base Directory spec
+  const xdgConfigHome = process.env.XDG_CONFIG_HOME || path.join(os.homedir(), ".config");
+  const configPath = path.join(xdgConfigHome, "opencode", "config.toml");
+  const configDir = path.dirname(configPath);
+
+  // Ensure ~/.config/opencode/ exists
+  await fs.mkdir(configDir, { recursive: true });
+
+  const normalizedBaseUrl = String(baseUrl || "")
+    .trim()
+    .replace(/\/+$/, "");
+
+  // Read existing TOML to preserve any user settings outside our block
+  let existingContent = "";
+  try {
+    existingContent = await fs.readFile(configPath, "utf-8");
+  } catch {
+    // File doesn't exist yet — start fresh
+  }
+
+  // Build the OmniRoute TOML block.
+  // opencode config.toml uses the [provider.X] table format.
+  void apiPort; // available for future port-based detection
+  const omniBlock = `
+# OmniRoute managed — updated automatically by OmniRoute CLI Tools
+[provider.omniroute]
+api_key = "${apiKey || "sk_omniroute"}"
+base_url = "${normalizedBaseUrl}"
+model = "${model}"
+`;
+
+  // Remove old OmniRoute-managed block (if any) then append fresh one
+  const cleanedContent = existingContent
+    .replace(/\n?# OmniRoute managed[\s\S]*?(?=\n\[|$)/, "")
+    .trimEnd();
+
+  const newContent = (cleanedContent ? cleanedContent + "\n" : "") + omniBlock;
+
+  await fs.writeFile(configPath, newContent, "utf-8");
+
+  return NextResponse.json({
+    success: true,
+    message: `OpenCode config saved to ${configPath}`,
+    configPath,
+  });
+}

src/app/api/cli-tools/openclaw-settings/route.ts

@@ -12,6 +12,7 @@ import { createBackup } from "@/shared/services/backupService";
 import { saveCliToolLastConfigured, deleteCliToolLastConfigured } from "@/lib/db/cliToolState";
 import { cliModelConfigSchema } from "@/shared/validation/schemas";
 import { isValidationFailure, validateBody } from "@/shared/validation/helpers";
+import { getApiKeyById } from "@/lib/localDb";
 
 const getOpenClawSettingsPath = () => getCliPrimaryConfigPath("openclaw");
 const getOpenClawDir = () => path.dirname(getOpenClawSettingsPath());
@@ -101,7 +102,18 @@ export async function POST(request: Request) {
     if (isValidationFailure(validation)) {
       return NextResponse.json({ error: validation.error }, { status: 400 });
     }
-    const { baseUrl, apiKey, model } = validation.data;
+    let { baseUrl, apiKey, model } = validation.data;
+
+    // (#526) Resolve real key from DB if keyId was provided
+    const keyId = typeof rawBody?.keyId === "string" ? rawBody.keyId.trim() : null;
+    if (keyId) {
+      try {
+        const keyRecord = await getApiKeyById(keyId);
+        if (keyRecord?.key) apiKey = keyRecord.key as string;
+      } catch {
+        /* non-critical */
+      }
+    }
 
     const openclawDir = getOpenClawDir();
     const settingsPath = getOpenClawSettingsPath();

src/app/api/provider-models/route.ts

@@ -6,6 +6,7 @@ import {
   updateCustomModel,
   getModelCompatOverrides,
   mergeModelCompatOverride,
+  type ModelCompatPatch,
 } from "@/lib/localDb";
 import {
   AI_PROVIDERS,
@@ -129,6 +130,7 @@ export async function PUT(request) {
       supportedEndpoints,
       normalizeToolCallId,
       preserveOpenAIDeveloperRole,
+      compatByProtocol,
     } = validation.data;
 
     const raw = rawBody as Record<string, unknown>;
@@ -139,6 +141,9 @@ export async function PUT(request) {
     if ("normalizeToolCallId" in raw) updates.normalizeToolCallId = normalizeToolCallId;
     if ("preserveOpenAIDeveloperRole" in raw)
       updates.preserveOpenAIDeveloperRole = preserveOpenAIDeveloperRole;
+    if ("compatByProtocol" in raw && compatByProtocol !== undefined) {
+      updates.compatByProtocol = compatByProtocol;
+    }
 
     const model = await updateCustomModel(provider, modelId, updates);
 
@@ -147,9 +152,17 @@ export async function PUT(request) {
       const compatOnly =
         rawKeys.length > 0 &&
         rawKeys.every((k) =>
-          ["provider", "modelId", "normalizeToolCallId", "preserveOpenAIDeveloperRole"].includes(k)
+          [
+            "provider",
+            "modelId",
+            "normalizeToolCallId",
+            "preserveOpenAIDeveloperRole",
+            "compatByProtocol",
+          ].includes(k)
         ) &&
-        ("normalizeToolCallId" in raw || "preserveOpenAIDeveloperRole" in raw);
+        ("normalizeToolCallId" in raw ||
+          "preserveOpenAIDeveloperRole" in raw ||
+          "compatByProtocol" in raw);
       if (compatOnly) {
         const knownProvider =
           !!provider &&
@@ -165,10 +178,7 @@ export async function PUT(request) {
             { status: 400 }
           );
         }
-        const patch: {
-          normalizeToolCallId?: boolean;
-          preserveOpenAIDeveloperRole?: boolean | null;
-        } = {};
+        const patch: ModelCompatPatch = {};
         if ("normalizeToolCallId" in raw && typeof normalizeToolCallId === "boolean") {
           patch.normalizeToolCallId = normalizeToolCallId;
         }
@@ -178,6 +188,9 @@ export async function PUT(request) {
               ? preserveOpenAIDeveloperRole
               : undefined;
         }
+        if ("compatByProtocol" in raw && compatByProtocol && typeof compatByProtocol === "object") {
+          patch.compatByProtocol = compatByProtocol;
+        }
         if (Object.keys(patch).length > 0) {
           mergeModelCompatOverride(provider, modelId, patch);
         }

src/app/api/sync/initialize/route.ts

@@ -1,7 +1,9 @@
 import { NextResponse } from "next/server";
 import initializeCloudSync from "@/shared/services/initializeCloudSync";
+import { startModelSyncScheduler } from "@/shared/services/modelSyncScheduler";
 
 let syncInitialized = false;
+let modelSyncInitialized = false;
 
 // POST /api/sync/initialize - Initialize cloud sync scheduler
 export async function POST(request) {
@@ -15,9 +17,17 @@ export async function POST(request) {
     await initializeCloudSync();
     syncInitialized = true;
 
+    // (#488) Start model auto-sync scheduler (24h, configurable via MODEL_SYNC_INTERVAL_HOURS)
+    if (!modelSyncInitialized) {
+      const origin = request.headers.get("origin") || "http://localhost:20128";
+      startModelSyncScheduler(origin);
+      modelSyncInitialized = true;
+    }
+
     return NextResponse.json({
       success: true,
       message: "Cloud sync initialized successfully",
+      modelSyncEnabled: true,
     });
   } catch (error) {
     console.log("Error initializing cloud sync:", error);
@@ -34,6 +44,7 @@ export async function POST(request) {
 export async function GET(request) {
   return NextResponse.json({
     initialized: syncInitialized,
+    modelSyncInitialized,
     message: syncInitialized ? "Cloud sync is running" : "Cloud sync not initialized",
   });
 }

src/app/api/usage/[connectionId]/route.ts

@@ -1,4 +1,8 @@
-import { getProviderConnectionById, updateProviderConnection, resolveProxyForConnection } from "@/lib/localDb";
+import {
+  getProviderConnectionById,
+  updateProviderConnection,
+  resolveProxyForConnection,
+} from "@/lib/localDb";
 import { getMachineId } from "@/shared/utils/machine";
 import { getUsageForProvider } from "@omniroute/open-sse/services/usage.ts";
 import { getExecutor } from "@omniroute/open-sse/executors/index.ts";
@@ -109,7 +113,10 @@ async function refreshAndUpdateCredentials(connection: any) {
 /**
  * GET /api/usage/[connectionId] - Get usage data for a specific connection
  */
-export async function GET(request: Request, { params }: { params: Promise<{ connectionId: string }> }) {
+export async function GET(
+  request: Request,
+  { params }: { params: Promise<{ connectionId: string }> }
+) {
   try {
     const { connectionId } = await params;
 
@@ -155,7 +162,34 @@ export async function GET(request: Request, { params }: { params: Promise<{ conn
 
     // Populate quota cache for quota-aware account selection
     if (isRecord(usage?.quotas)) {
-      setQuotaCache(connectionId, connection.provider, usage.quotas);
+      setQuotaCache(
+        connectionId,
+        connection.provider as string,
+        usage.quotas as Record<string, unknown>
+      );
+    }
+
+    // (#491) If the live usage check returned an auth error, sync the expired status
+    // back to the DB so the Providers page reflects the same degraded state as
+    // Limits & Quotas (which performs the live check).
+    const errorMessage = typeof usage?.message === "string" ? usage.message.toLowerCase() : "";
+    const isAuthError =
+      errorMessage.includes("token expired") ||
+      errorMessage.includes("access denied") ||
+      errorMessage.includes("re-authenticate") ||
+      errorMessage.includes("unauthorized");
+
+    if (isAuthError && connection.testStatus !== "expired") {
+      try {
+        await updateProviderConnection(connection.id as string, {
+          testStatus: "expired",
+          lastErrorType: "token_expired",
+          lastErrorAt: new Date().toISOString(),
+        });
+      } catch (dbErr) {
+        // Non-critical: log but don't block the response
+        console.error("[Usage API] Failed to sync expired status to DB:", dbErr);
+      }
     }
 
     return Response.json(usage);

src/app/api/v1/accounts/[id]/limits/route.ts

@@ -0,0 +1,49 @@
+import { NextResponse } from "next/server";
+import { z } from "zod";
+import { isAuthenticated } from "@/shared/utils/apiAuth";
+import { setAccountKeyLimit, getAccountKeyLimit } from "@/lib/db/registeredKeys";
+
+const limitsSchema = z.object({
+  maxActiveKeys: z.number().int().positive().nullable().optional(),
+  dailyIssueLimit: z.number().int().positive().nullable().optional(),
+  hourlyIssueLimit: z.number().int().positive().nullable().optional(),
+});
+
+/**
+ * GET /api/v1/accounts/[id]/limits
+ * Get the current issuance limits for an account.
+ */
+export async function GET(request: Request, { params }: { params: { id: string } }) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  const limits = getAccountKeyLimit(params.id);
+  return NextResponse.json({ accountId: params.id, limits: limits ?? null });
+}
+
+/**
+ * PUT /api/v1/accounts/[id]/limits
+ * Configure issuance limits for an account.
+ */
+export async function PUT(request: Request, { params }: { params: { id: string } }) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  let body: unknown;
+  try {
+    body = await request.json();
+  } catch {
+    return NextResponse.json({ error: "Invalid JSON body" }, { status: 400 });
+  }
+
+  const parsed = limitsSchema.safeParse(body);
+  if (!parsed.success) {
+    return NextResponse.json({ error: parsed.error.flatten() }, { status: 400 });
+  }
+
+  setAccountKeyLimit(params.id, parsed.data);
+  const updated = getAccountKeyLimit(params.id);
+  return NextResponse.json({ accountId: params.id, limits: updated });
+}

src/app/api/v1/embeddings/route.ts

@@ -160,6 +160,7 @@ export async function POST(request) {
   // Resolve provider config — dynamic first (local override), then hardcoded
   let providerConfig: EmbeddingProvider | null =
     dynamicProviders.find((dp) => dp.id === provider) || getEmbeddingProvider(provider) || null;
+  let credentialsProviderId = provider;
 
   // #496: Fallback — resolve from ALL provider_nodes (not just localhost)
   // This enables custom embedding models (e.g. google/gemini-embedding-001) whose
@@ -180,6 +181,7 @@ export async function POST(request) {
           authHeader: "bearer",
           models: [],
         };
+        credentialsProviderId = matchingNode.id || provider;
         log.info(
           "EMBED",
           `Resolved custom embedding provider: ${provider} → ${providerConfig.baseUrl}`
@@ -200,7 +202,7 @@ export async function POST(request) {
   // Get credentials — skip for local providers (authType: "none")
   let credentials = null;
   if (providerConfig && providerConfig.authType !== "none") {
-    credentials = await getProviderCredentials(provider);
+    credentials = await getProviderCredentials(credentialsProviderId);
     if (!credentials) {
       return errorResponse(
         HTTP_STATUS.BAD_REQUEST,

src/app/api/v1/issues/report/route.ts

@@ -0,0 +1,123 @@
+import { NextResponse } from "next/server";
+import { z } from "zod";
+import { isAuthenticated } from "@/shared/utils/apiAuth";
+
+const reportSchema = z.object({
+  title: z.string().min(1).max(300),
+  provider: z.string().max(80).optional(),
+  accountId: z.string().max(120).optional(),
+  requestId: z.string().max(200).optional(),
+  errorCode: z.string().max(100).optional(),
+  details: z.record(z.unknown()).optional(),
+  labels: z.array(z.string().max(50)).optional(),
+});
+
+/**
+ * POST /api/v1/issues/report
+ *
+ * Optionally report a quota-exceeded or key-issuance failure event to GitHub.
+ *
+ * Requires GITHUB_ISSUES_REPO (format: owner/repo) and GITHUB_ISSUES_TOKEN
+ * environment variables to be set. If not configured, returns 202 (accepted but
+ * logged only).
+ */
+export async function POST(request: Request) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  let body: unknown;
+  try {
+    body = await request.json();
+  } catch {
+    return NextResponse.json({ error: "Invalid JSON body" }, { status: 400 });
+  }
+
+  const parsed = reportSchema.safeParse(body);
+  if (!parsed.success) {
+    return NextResponse.json({ error: parsed.error.flatten() }, { status: 400 });
+  }
+
+  const { title, provider, accountId, requestId, errorCode, details, labels = [] } = parsed.data;
+
+  const repo = process.env.GITHUB_ISSUES_REPO;
+  const token = process.env.GITHUB_ISSUES_TOKEN;
+
+  // ── Structured body for the GitHub issue ──
+  const issueBody = [
+    `## ${errorCode ?? "Key Issuance Event"}`,
+    "",
+    "| Field | Value |",
+    "|-------|-------|",
+    provider ? `| Provider | \`${provider}\` |` : null,
+    accountId ? `| Account ID | \`${accountId}\` |` : null,
+    requestId ? `| Request ID | \`${requestId}\` |` : null,
+    errorCode ? `| Error Code | \`${errorCode}\` |` : null,
+    `| Reported At | ${new Date().toISOString()} |`,
+    "",
+    details ? "### Details\n```json\n" + JSON.stringify(details, null, 2) + "\n```" : null,
+    "",
+    "_Auto-reported by OmniRoute Registered Key Issuer_",
+  ]
+    .filter(Boolean)
+    .join("\n");
+
+  // ── Log locally regardless ──
+  console.log(
+    `[issues/report] title="${title}" errorCode=${errorCode ?? "—"} provider=${provider ?? "—"} accountId=${accountId ?? "—"}`
+  );
+
+  if (!repo || !token) {
+    // No GitHub config — log only
+    return NextResponse.json(
+      {
+        logged: true,
+        githubIssueCreated: false,
+        reason: !repo ? "GITHUB_ISSUES_REPO not configured" : "GITHUB_ISSUES_TOKEN not configured",
+      },
+      { status: 202 }
+    );
+  }
+
+  // ── Create GitHub issue ──
+  try {
+    const [owner, repoName] = repo.split("/");
+    const ghRes = await fetch(`https://api.github.com/repos/${owner}/${repoName}/issues`, {
+      method: "POST",
+      headers: {
+        Authorization: `Bearer ${token}`,
+        Accept: "application/vnd.github+json",
+        "X-GitHub-Api-Version": "2022-11-28",
+        "Content-Type": "application/json",
+      },
+      body: JSON.stringify({
+        title: `[Key Issuer] ${title}`,
+        body: issueBody,
+        labels: ["key-issuer", "automated", ...labels],
+      }),
+    });
+
+    if (!ghRes.ok) {
+      const errText = await ghRes.text();
+      console.error(`[issues/report] GitHub API error ${ghRes.status}: ${errText}`);
+      return NextResponse.json(
+        { logged: true, githubIssueCreated: false, githubError: ghRes.status },
+        { status: 207 }
+      );
+    }
+
+    const ghData = await ghRes.json();
+    return NextResponse.json({
+      logged: true,
+      githubIssueCreated: true,
+      githubIssueUrl: ghData.html_url,
+      githubIssueNumber: ghData.number,
+    });
+  } catch (err) {
+    console.error("[issues/report] GitHub fetch failed:", err);
+    return NextResponse.json(
+      { logged: true, githubIssueCreated: false, error: "GitHub request failed" },
+      { status: 207 }
+    );
+  }
+}

src/app/api/v1/providers/[id]/limits/route.ts

@@ -0,0 +1,49 @@
+import { NextResponse } from "next/server";
+import { z } from "zod";
+import { isAuthenticated } from "@/shared/utils/apiAuth";
+import { setProviderKeyLimit, getProviderKeyLimit } from "@/lib/db/registeredKeys";
+
+const limitsSchema = z.object({
+  maxActiveKeys: z.number().int().positive().nullable().optional(),
+  dailyIssueLimit: z.number().int().positive().nullable().optional(),
+  hourlyIssueLimit: z.number().int().positive().nullable().optional(),
+});
+
+/**
+ * GET /api/v1/providers/[id]/limits
+ * Get the current issuance limits for a provider.
+ */
+export async function GET(request: Request, { params }: { params: { id: string } }) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  const limits = getProviderKeyLimit(params.id);
+  return NextResponse.json({ provider: params.id, limits: limits ?? null });
+}
+
+/**
+ * PUT /api/v1/providers/[id]/limits
+ * Configure issuance limits for a provider.
+ */
+export async function PUT(request: Request, { params }: { params: { id: string } }) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  let body: unknown;
+  try {
+    body = await request.json();
+  } catch {
+    return NextResponse.json({ error: "Invalid JSON body" }, { status: 400 });
+  }
+
+  const parsed = limitsSchema.safeParse(body);
+  if (!parsed.success) {
+    return NextResponse.json({ error: parsed.error.flatten() }, { status: 400 });
+  }
+
+  setProviderKeyLimit(params.id, parsed.data);
+  const updated = getProviderKeyLimit(params.id);
+  return NextResponse.json({ provider: params.id, limits: updated });
+}

src/app/api/v1/quotas/check/route.ts

@@ -0,0 +1,33 @@
+import { NextResponse } from "next/server";
+import { isAuthenticated } from "@/shared/utils/apiAuth";
+import { checkQuota } from "@/lib/db/registeredKeys";
+
+/**
+ * GET /api/v1/quotas/check?provider=&accountId=
+ *
+ * Check if a new registered key can be issued for the given provider/account
+ * without actually issuing one. Use this to pre-validate before POST /registered-keys.
+ */
+export async function GET(request: Request) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  const { searchParams } = new URL(request.url);
+  const provider = searchParams.get("provider") ?? "";
+  const accountId = searchParams.get("accountId") ?? "";
+
+  try {
+    const result = checkQuota(provider, accountId);
+    return NextResponse.json({
+      allowed: result.allowed,
+      ...(result.errorCode ? { errorCode: result.errorCode, reason: result.errorMessage } : {}),
+      provider: provider || null,
+      accountId: accountId || null,
+      checkedAt: new Date().toISOString(),
+    });
+  } catch (err) {
+    console.error("[quotas/check] error:", err);
+    return NextResponse.json({ error: "Quota check failed" }, { status: 500 });
+  }
+}

src/app/api/v1/registered-keys/[id]/revoke/route.ts

@@ -0,0 +1,21 @@
+import { NextResponse } from "next/server";
+import { isAuthenticated } from "@/shared/utils/apiAuth";
+import { revokeRegisteredKey } from "@/lib/db/registeredKeys";
+
+/**
+ * POST /api/v1/registered-keys/[id]/revoke
+ *
+ * Explicit revoke endpoint (supports clients that cannot issue DELETE requests).
+ */
+export async function POST(request: Request, { params }: { params: { id: string } }) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  const revoked = revokeRegisteredKey(params.id);
+  if (!revoked) {
+    return NextResponse.json({ error: "Key not found or already revoked" }, { status: 404 });
+  }
+
+  return NextResponse.json({ success: true, id: params.id, revokedAt: new Date().toISOString() });
+}

src/app/api/v1/registered-keys/[id]/route.ts

@@ -0,0 +1,33 @@
+import { NextResponse } from "next/server";
+import { isAuthenticated } from "@/shared/utils/apiAuth";
+import { getRegisteredKey, revokeRegisteredKey } from "@/lib/db/registeredKeys";
+
+// ─── GET /api/v1/registered-keys/[id] ────────────────────────────────────────
+
+export async function GET(request: Request, { params }: { params: { id: string } }) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  const key = getRegisteredKey(params.id);
+  if (!key) {
+    return NextResponse.json({ error: "Key not found" }, { status: 404 });
+  }
+
+  return NextResponse.json({ key });
+}
+
+// ─── DELETE /api/v1/registered-keys/[id] ─────────────────────────────────────
+
+export async function DELETE(request: Request, { params }: { params: { id: string } }) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  const revoked = revokeRegisteredKey(params.id);
+  if (!revoked) {
+    return NextResponse.json({ error: "Key not found or already revoked" }, { status: 404 });
+  }
+
+  return NextResponse.json({ success: true, id: params.id, revokedAt: new Date().toISOString() });
+}

src/app/api/v1/registered-keys/route.ts

@@ -0,0 +1,118 @@
+import { NextResponse } from "next/server";
+import { z } from "zod";
+import { isAuthenticated } from "@/shared/utils/apiAuth";
+import { issueRegisteredKey, checkQuota, listRegisteredKeys } from "@/lib/db/registeredKeys";
+
+// ─── Validation ───────────────────────────────────────────────────────────────
+
+const issueKeySchema = z.object({
+  name: z.string().min(1).max(120),
+  provider: z.string().max(80).optional().default(""),
+  accountId: z.string().max(120).optional().default(""),
+  idempotencyKey: z.string().max(256).optional(),
+  expiresAt: z.string().datetime({ offset: true }).optional(),
+  dailyBudget: z.number().int().positive().optional(),
+  hourlyBudget: z.number().int().positive().optional(),
+});
+
+// ─── GET /api/v1/registered-keys ─────────────────────────────────────────────
+
+/**
+ * List registered keys (masked — no raw key material returned after creation).
+ * Optional query params: ?provider=&accountId=
+ */
+export async function GET(request: Request) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  const { searchParams } = new URL(request.url);
+  const provider = searchParams.get("provider") ?? undefined;
+  const accountId = searchParams.get("accountId") ?? undefined;
+
+  try {
+    const keys = listRegisteredKeys({ provider, accountId });
+    return NextResponse.json({ keys, total: keys.length });
+  } catch (err) {
+    console.error("[registered-keys] GET failed:", err);
+    return NextResponse.json({ error: "Failed to list registered keys" }, { status: 500 });
+  }
+}
+
+// ─── POST /api/v1/registered-keys ────────────────────────────────────────────
+
+/**
+ * Issue a new registered key.
+ *
+ * Checks provider + account quotas before issuing.
+ * Returns the raw key ONCE — it is never stored in plain text.
+ * Subsequent fetches will only return the masked prefix.
+ */
+export async function POST(request: Request) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: { message: "Authentication required" } }, { status: 401 });
+  }
+
+  let body: unknown;
+  try {
+    body = await request.json();
+  } catch {
+    return NextResponse.json({ error: "Invalid JSON body" }, { status: 400 });
+  }
+
+  const parsed = issueKeySchema.safeParse(body);
+  if (!parsed.success) {
+    return NextResponse.json({ error: parsed.error.flatten() }, { status: 400 });
+  }
+
+  const { provider, accountId } = parsed.data;
+
+  // ── Quota check ──
+  try {
+    const quota = checkQuota(provider, accountId);
+    if (!quota.allowed) {
+      return NextResponse.json(
+        { error: quota.errorMessage, errorCode: quota.errorCode },
+        { status: 429 }
+      );
+    }
+  } catch (err) {
+    console.error("[registered-keys] quota check failed:", err);
+    return NextResponse.json({ error: "Quota check failed" }, { status: 500 });
+  }
+
+  // ── Issue ──
+  try {
+    const result = issueRegisteredKey(parsed.data);
+
+    if ("idempotencyConflict" in result) {
+      return NextResponse.json(
+        {
+          error: "Idempotency key already used",
+          errorCode: "IDEMPOTENCY_CONFLICT",
+          existing: result.existing,
+        },
+        { status: 409 }
+      );
+    }
+
+    const { rawKey, ...keyMeta } = result;
+    return NextResponse.json(
+      {
+        key: rawKey, // ← shown ONCE only
+        keyId: keyMeta.id,
+        keyPrefix: keyMeta.keyPrefix,
+        name: keyMeta.name,
+        provider: keyMeta.provider,
+        accountId: keyMeta.accountId,
+        expiresAt: keyMeta.expiresAt,
+        createdAt: keyMeta.createdAt,
+        warning: "Store this key securely — it will not be shown again.",
+      },
+      { status: 201 }
+    );
+  } catch (err) {
+    console.error("[registered-keys] issue failed:", err);
+    return NextResponse.json({ error: "Failed to issue key" }, { status: 500 });
+  }
+}

src/app/login/page.tsx

@@ -69,6 +69,11 @@ export default function LoginPage() {
         router.refresh();
       } else {
         const data = await res.json();
+        // (#521) If no password is set, redirect to onboarding instead of showing an error
+        if (data.needsSetup) {
+          router.push("/dashboard/onboarding");
+          return;
+        }
         setError(data.error || t("invalidPassword"));
       }
     } catch (err) {

src/i18n/messages/en.json

@@ -281,6 +281,7 @@
     "failedUpdatePermissionsRetry": "Failed to update permissions. Please try again.",
     "unknownProvider": "unknown",
     "copyMaskedKey": "Copy masked key",
+    "keyOnlyAvailableAtCreation": "Full key available only at creation time — copy it when you first create the key",
     "modelsCount": "{count, plural, one {# model} other {# models}}",
     "lastUsedOn": "Last: {date}",
     "editPermissions": "Edit permissions",
@@ -1437,6 +1438,11 @@
     "compatDeveloperShort": "Developer role",
     "compatDoNotPreserveDeveloper": "Do not preserve developer role",
     "compatBadgeNoPreserve": "No preserve",
+    "compatProtocolLabel": "Client request protocol",
+    "compatProtocolHint": "These options apply when OmniRoute detects this request shape (OpenAI Chat, Responses API, or Anthropic Messages).",
+    "compatProtocolOpenAI": "OpenAI Chat Completions",
+    "compatProtocolOpenAIResponses": "OpenAI Responses API",
+    "compatProtocolClaude": "Anthropic Messages",
     "modelId": "Model ID",
     "customModelPlaceholder": "e.g. gpt-4.5-turbo",
     "loading": "Loading...",

src/i18n/messages/zh-CN.json

@@ -1432,6 +1432,11 @@
     "compatDeveloperShort": "Developer 角色",
     "compatDoNotPreserveDeveloper": "不保留 developer 角色",
     "compatBadgeNoPreserve": "不保留",
+    "compatProtocolLabel": "客户端请求协议",
+    "compatProtocolHint": "以下选项在 OmniRoute 识别到该请求形态（OpenAI Chat、Responses API 或 Anthropic Messages）时生效。",
+    "compatProtocolOpenAI": "OpenAI Chat Completions",
+    "compatProtocolOpenAIResponses": "OpenAI Responses API",
+    "compatProtocolClaude": "Anthropic Messages",
     "modelId": "模型 ID",
     "customModelPlaceholder": "例如：gpt-4.5-turbo",
     "loading": "正在加载...",

src/lib/db/apiKeys.ts

@@ -40,6 +40,8 @@ interface ApiKeyMetadata {
   accessSchedule: AccessSchedule | null;
   maxRequestsPerDay: number | null;
   maxRequestsPerMinute: number | null;
+  // T08: Per-key max concurrent sticky sessions (0 = unlimited)
+  maxSessions: number;
 }
 
 interface ApiKeyRow extends JsonRecord {
@@ -197,6 +199,11 @@ function ensureApiKeysColumns(db: ApiKeysDbLike) {
       db.exec("ALTER TABLE api_keys ADD COLUMN max_requests_per_minute INTEGER");
       console.log("[DB] Added api_keys.max_requests_per_minute column");
     }
+    // T08: max concurrent sticky sessions per key (0 = unlimited)
+    if (!columnNames.has("max_sessions")) {
+      db.exec("ALTER TABLE api_keys ADD COLUMN max_sessions INTEGER NOT NULL DEFAULT 0");
+      console.log("[DB] Added api_keys.max_sessions column");
+    }
     _schemaChecked = true;
   } catch (error) {
     const message = error instanceof Error ? error.message : String(error);
@@ -222,7 +229,7 @@ function getPreparedStatements(db: ApiKeysDbLike): ApiKeysStatements {
     _stmtGetKeyById = db.prepare<ApiKeyRow>("SELECT * FROM api_keys WHERE id = ?");
     _stmtValidateKey = db.prepare<JsonRecord>("SELECT 1 FROM api_keys WHERE key = ?");
     _stmtGetKeyMetadata = db.prepare<ApiKeyRow>(
-      "SELECT id, name, machine_id, allowed_models, allowed_connections, no_log, auto_resolve, is_active, access_schedule, max_requests_per_day, max_requests_per_minute FROM api_keys WHERE key = ?"
+      "SELECT id, name, machine_id, allowed_models, allowed_connections, no_log, auto_resolve, is_active, access_schedule, max_requests_per_day, max_requests_per_minute, max_sessions FROM api_keys WHERE key = ?"
     );
     _stmtInsertKey = db.prepare(
       "INSERT INTO api_keys (id, name, key, machine_id, allowed_models, no_log, created_at) VALUES (?, ?, ?, ?, ?, ?, ?)"
@@ -418,6 +425,8 @@ export async function updateApiKeyPermissions(
         accessSchedule?: AccessSchedule | null;
         maxRequestsPerDay?: number | null;
         maxRequestsPerMinute?: number | null;
+        // T08: max concurrent sessions for this key (0 = unlimited)
+        maxSessions?: number | null;
       }
 ) {
   const db = getDbInstance() as ApiKeysDbLike;
@@ -436,6 +445,7 @@ export async function updateApiKeyPermissions(
           accessSchedule: update.accessSchedule,
           maxRequestsPerDay: update.maxRequestsPerDay,
           maxRequestsPerMinute: update.maxRequestsPerMinute,
+          maxSessions: (update as { maxSessions?: number | null }).maxSessions,
         };
 
   if (
@@ -447,7 +457,8 @@ export async function updateApiKeyPermissions(
     normalized.isActive === undefined &&
     normalized.accessSchedule === undefined &&
     normalized.maxRequestsPerDay === undefined &&
-    normalized.maxRequestsPerMinute === undefined
+    normalized.maxRequestsPerMinute === undefined &&
+    (normalized as Record<string, unknown>).maxSessions === undefined
   ) {
     return false;
   }
@@ -464,6 +475,7 @@ export async function updateApiKeyPermissions(
     accessSchedule?: string | null;
     maxRequestsPerDay?: number | null;
     maxRequestsPerMinute?: number | null;
+    maxSessions?: number;
   } = { id };
 
   if (normalized.name !== undefined) {
@@ -514,6 +526,12 @@ export async function updateApiKeyPermissions(
     params.maxRequestsPerMinute = normalized.maxRequestsPerMinute;
   }
 
+  const maxSessionsUpdate = (normalized as Record<string, unknown>).maxSessions;
+  if (maxSessionsUpdate !== undefined) {
+    updates.push("max_sessions = @maxSessions");
+    params.maxSessions = typeof maxSessionsUpdate === "number" ? Math.max(0, maxSessionsUpdate) : 0;
+  }
+
   const result = db.prepare(`UPDATE api_keys SET ${updates.join(", ")} WHERE id = @id`).run(params);
 
   if (result.changes === 0) return false;
@@ -605,6 +623,8 @@ export async function getApiKeyMetadata(
   const rawMaxRPD = record.max_requests_per_day ?? record.maxRequestsPerDay;
   const rawMaxRPM = record.max_requests_per_minute ?? record.maxRequestsPerMinute;
 
+  const rawMaxSessions = record.max_sessions ?? record.maxSessions;
+
   const metadata: ApiKeyMetadata = {
     id: metadataId,
     name: metadataName,
@@ -619,6 +639,8 @@ export async function getApiKeyMetadata(
     accessSchedule: parseAccessSchedule(record.access_schedule ?? record.accessSchedule),
     maxRequestsPerDay: typeof rawMaxRPD === "number" && rawMaxRPD > 0 ? rawMaxRPD : null,
     maxRequestsPerMinute: typeof rawMaxRPM === "number" && rawMaxRPM > 0 ? rawMaxRPM : null,
+    // T08: max concurrent sessions; 0 = unlimited (default & backward-compatible)
+    maxSessions: typeof rawMaxSessions === "number" && rawMaxSessions > 0 ? rawMaxSessions : 0,
   };
 
   if (!metadata.id) {

src/lib/db/migrations/008_registered_keys.sql

@@ -0,0 +1,65 @@
+-- Migration 008: Registered Keys Provisioning API (#464)
+--
+-- Adds three tables:
+--   registered_keys      — auto-provisioned API keys with quota metadata
+--   provider_key_limits  — per-provider issuance limits
+--   account_key_limits   — per-account issuance limits
+
+-- --------------------------------------------------------------------------
+-- Table: registered_keys
+-- --------------------------------------------------------------------------
+CREATE TABLE IF NOT EXISTS registered_keys (
+  id            TEXT PRIMARY KEY,           -- UUID
+  key           TEXT NOT NULL UNIQUE,       -- hashed key material (sha256)
+  key_prefix    TEXT NOT NULL,              -- first 8 chars for display (e.g. "ork_abc1")
+  name          TEXT NOT NULL,
+  provider      TEXT NOT NULL DEFAULT '',   -- associated provider (optional)
+  account_id    TEXT NOT NULL DEFAULT '',   -- account/tenant identifier
+  is_active     INTEGER NOT NULL DEFAULT 1,
+  revoked_at    TEXT,                       -- ISO timestamp, null if active
+  expires_at    TEXT,                       -- ISO timestamp, null = no expiry
+  idempotency_key TEXT UNIQUE,             -- prevents duplicate issue requests
+  daily_budget  INTEGER,                   -- max requests per day (null = unlimited)
+  hourly_budget INTEGER,                   -- max requests per hour (null = unlimited)
+  daily_used    INTEGER NOT NULL DEFAULT 0,
+  hourly_used   INTEGER NOT NULL DEFAULT 0,
+  last_reset_day  TEXT NOT NULL DEFAULT '',  -- YYYY-MM-DD for daily reset tracking
+  last_reset_hour TEXT NOT NULL DEFAULT '',  -- YYYY-MM-DDTHH for hourly reset tracking
+  created_at    TEXT NOT NULL DEFAULT (datetime('now')),
+  updated_at    TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_registered_keys_provider  ON registered_keys(provider);
+CREATE INDEX IF NOT EXISTS idx_registered_keys_account   ON registered_keys(account_id);
+CREATE INDEX IF NOT EXISTS idx_registered_keys_active    ON registered_keys(is_active);
+CREATE INDEX IF NOT EXISTS idx_registered_keys_idempotency ON registered_keys(idempotency_key);
+
+-- --------------------------------------------------------------------------
+-- Table: provider_key_limits  (per-provider issuance limits)
+-- --------------------------------------------------------------------------
+CREATE TABLE IF NOT EXISTS provider_key_limits (
+  provider         TEXT PRIMARY KEY,
+  max_active_keys  INTEGER,   -- null = unlimited
+  daily_issue_limit  INTEGER, -- max keys per day
+  hourly_issue_limit INTEGER, -- max keys per hour
+  daily_issued     INTEGER NOT NULL DEFAULT 0,
+  hourly_issued    INTEGER NOT NULL DEFAULT 0,
+  last_reset_day   TEXT NOT NULL DEFAULT '',
+  last_reset_hour  TEXT NOT NULL DEFAULT '',
+  updated_at       TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+-- --------------------------------------------------------------------------
+-- Table: account_key_limits  (per-account issuance limits)
+-- --------------------------------------------------------------------------
+CREATE TABLE IF NOT EXISTS account_key_limits (
+  account_id       TEXT PRIMARY KEY,
+  max_active_keys  INTEGER,
+  daily_issue_limit  INTEGER,
+  hourly_issue_limit INTEGER,
+  daily_issued     INTEGER NOT NULL DEFAULT 0,
+  hourly_issued    INTEGER NOT NULL DEFAULT 0,
+  last_reset_day   TEXT NOT NULL DEFAULT '',
+  last_reset_hour  TEXT NOT NULL DEFAULT '',
+  updated_at       TEXT NOT NULL DEFAULT (datetime('now'))
+);

src/lib/db/migrations/009_requested_model.sql

@@ -0,0 +1,9 @@
+-- Migration 009: Add requested_model to call_logs for billing transparency
+-- Tracks the model the client *asked* for vs the model that was *actually routed*.
+-- Needed when a combo falls back: requested_model ≠ model in call_logs.
+-- Ref: sub2api commits 0b845c25 + 4edcfe1f (T01 sub2api gap analysis)
+ALTER TABLE call_logs ADD COLUMN requested_model TEXT DEFAULT NULL;
+
+-- Index for filtering/aggregating by requested_model in Analytics
+CREATE INDEX IF NOT EXISTS idx_call_logs_requested_model
+  ON call_logs(requested_model);

src/lib/db/models.ts

@@ -4,16 +4,60 @@
 
 import { getDbInstance } from "./core";
 import { backupDbFile } from "./backup";
+import {
+  MODEL_COMPAT_PROTOCOL_KEYS,
+  type ModelCompatProtocolKey,
+} from "@/shared/constants/modelCompat";
 
 type JsonRecord = Record<string, unknown>;
 
 /** Built-in / alias models: tool-call + developer-role flags without a full custom row */
 const MODEL_COMPAT_NAMESPACE = "modelCompatOverrides";
 
+export { MODEL_COMPAT_PROTOCOL_KEYS, type ModelCompatProtocolKey };
+
+export type ModelCompatPerProtocol = {
+  normalizeToolCallId?: boolean;
+  preserveOpenAIDeveloperRole?: boolean;
+};
+
+type CompatByProtocolMap = Partial<Record<ModelCompatProtocolKey, ModelCompatPerProtocol>>;
+
+function isCompatProtocolKey(p: string): p is ModelCompatProtocolKey {
+  return (MODEL_COMPAT_PROTOCOL_KEYS as readonly string[]).includes(p);
+}
+
+function deepMergeCompatByProtocol(
+  prev: CompatByProtocolMap | undefined,
+  patch: Partial<Record<ModelCompatProtocolKey, Partial<ModelCompatPerProtocol>>>
+): CompatByProtocolMap {
+  const out: CompatByProtocolMap = { ...(prev || {}) };
+  for (const key of Object.keys(patch) as ModelCompatProtocolKey[]) {
+    if (!isCompatProtocolKey(key)) continue;
+    const deltas = patch[key];
+    if (!deltas || typeof deltas !== "object") continue;
+    const hasDelta =
+      Object.prototype.hasOwnProperty.call(deltas, "normalizeToolCallId") ||
+      Object.prototype.hasOwnProperty.call(deltas, "preserveOpenAIDeveloperRole");
+    if (!hasDelta) continue;
+    const cur: ModelCompatPerProtocol = { ...(out[key] || {}) };
+    if ("normalizeToolCallId" in deltas) {
+      cur.normalizeToolCallId = Boolean(deltas.normalizeToolCallId);
+    }
+    if ("preserveOpenAIDeveloperRole" in deltas) {
+      cur.preserveOpenAIDeveloperRole = Boolean(deltas.preserveOpenAIDeveloperRole);
+    }
+    if (Object.keys(cur).length === 0) delete out[key];
+    else out[key] = cur;
+  }
+  return out;
+}
+
 export type ModelCompatOverride = {
   id: string;
   normalizeToolCallId?: boolean;
   preserveOpenAIDeveloperRole?: boolean;
+  compatByProtocol?: CompatByProtocolMap;
 };
 
 function readCompatList(providerId: string): ModelCompatOverride[] {
@@ -52,13 +96,24 @@ export function getModelCompatOverrides(providerId: string): ModelCompatOverride
   return readCompatList(providerId);
 }
 
+export type ModelCompatPatch = {
+  normalizeToolCallId?: boolean;
+  preserveOpenAIDeveloperRole?: boolean | null;
+  compatByProtocol?: CompatByProtocolMap;
+};
+
+function compatByProtocolHasEntries(map: CompatByProtocolMap | undefined): boolean {
+  if (!map || typeof map !== "object") return false;
+  return Object.keys(map).some((k) => {
+    const v = map[k as ModelCompatProtocolKey];
+    return v && typeof v === "object" && Object.keys(v).length > 0;
+  });
+}
+
 export function mergeModelCompatOverride(
   providerId: string,
   modelId: string,
-  patch: Partial<{
-    normalizeToolCallId: boolean;
-    preserveOpenAIDeveloperRole: boolean | null;
-  }>
+  patch: ModelCompatPatch
 ) {
   const list = readCompatList(providerId);
   const idx = list.findIndex((e) => e.id === modelId);
@@ -75,9 +130,18 @@ export function mergeModelCompatOverride(
       next.preserveOpenAIDeveloperRole = Boolean(patch.preserveOpenAIDeveloperRole);
     }
   }
+  if (patch.compatByProtocol && Object.keys(patch.compatByProtocol).length > 0) {
+    const merged = deepMergeCompatByProtocol(next.compatByProtocol, patch.compatByProtocol);
+    if (compatByProtocolHasEntries(merged)) next.compatByProtocol = merged;
+    else delete next.compatByProtocol;
+  }
   const filtered = list.filter((e) => e.id !== modelId);
   const hasPreserveFlag = Object.prototype.hasOwnProperty.call(next, "preserveOpenAIDeveloperRole");
-  if (next.normalizeToolCallId || hasPreserveFlag) {
+  if (
+    next.normalizeToolCallId ||
+    hasPreserveFlag ||
+    compatByProtocolHasEntries(next.compatByProtocol)
+  ) {
     filtered.push(next);
   }
   writeCompatList(providerId, filtered);
@@ -281,6 +345,23 @@ export async function updateCustomModel(
   if (index === -1) return null;
 
   const current = models[index];
+  const currentCompat = (current as JsonRecord).compatByProtocol as CompatByProtocolMap | undefined;
+  let mergedCompat: CompatByProtocolMap | undefined = currentCompat;
+  if (
+    updates.compatByProtocol !== undefined &&
+    typeof updates.compatByProtocol === "object" &&
+    updates.compatByProtocol !== null &&
+    !Array.isArray(updates.compatByProtocol)
+  ) {
+    mergedCompat = deepMergeCompatByProtocol(
+      currentCompat,
+      updates.compatByProtocol as Partial<
+        Record<ModelCompatProtocolKey, Partial<ModelCompatPerProtocol>>
+      >
+    );
+    if (!compatByProtocolHasEntries(mergedCompat)) mergedCompat = undefined;
+  }
+
   const next: JsonRecord = {
     ...current,
     ...(updates.modelName !== undefined ? { name: updates.modelName || current.name } : {}),
@@ -299,6 +380,13 @@ export async function updateCustomModel(
       next.preserveOpenAIDeveloperRole = Boolean(updates.preserveOpenAIDeveloperRole);
     }
   }
+  if (updates.compatByProtocol !== undefined) {
+    if (mergedCompat && compatByProtocolHasEntries(mergedCompat)) {
+      next.compatByProtocol = mergedCompat;
+    } else {
+      delete next.compatByProtocol;
+    }
+  }
 
   models[index] = next;
 
@@ -335,11 +423,33 @@ function getCustomModelRow(providerId: string, modelId: string): JsonRecord | nu
 /**
  * Whether the given provider/model has "normalize tool call id" (9-char Mistral-style) enabled.
  * Custom model row wins; otherwise {@link getModelCompatOverrides}.
+ * When `sourceFormat` is one of `openai` | `openai-responses` | `claude`, per-protocol
+ * `compatByProtocol[sourceFormat].normalizeToolCallId` overrides the legacy top-level flag.
  */
-export function getModelNormalizeToolCallId(providerId: string, modelId: string): boolean {
+export function getModelNormalizeToolCallId(
+  providerId: string,
+  modelId: string,
+  sourceFormat?: string | null
+): boolean {
   const m = getCustomModelRow(providerId, modelId);
-  if (m) return Boolean(m.normalizeToolCallId);
+  const protocol = sourceFormat && isCompatProtocolKey(sourceFormat) ? sourceFormat : null;
+
+  if (m) {
+    if (protocol) {
+      const pc = (m.compatByProtocol as CompatByProtocolMap | undefined)?.[protocol];
+      if (pc && Object.prototype.hasOwnProperty.call(pc, "normalizeToolCallId")) {
+        return Boolean(pc.normalizeToolCallId);
+      }
+    }
+    return Boolean(m.normalizeToolCallId);
+  }
   const co = readCompatList(providerId).find((e) => e.id === modelId);
+  if (protocol && co?.compatByProtocol?.[protocol]) {
+    const pc = co.compatByProtocol[protocol]!;
+    if (Object.prototype.hasOwnProperty.call(pc, "normalizeToolCallId")) {
+      return Boolean(pc.normalizeToolCallId);
+    }
+  }
   return Boolean(co?.normalizeToolCallId);
 }
 
@@ -347,19 +457,35 @@ export function getModelNormalizeToolCallId(providerId: string, modelId: string)
  * Explicit preserve-openai-developer preference for this provider/model.
  * `undefined` = unset → routing keeps legacy default (preserve developer for OpenAI format).
  * `false` = map developer → system (e.g. MiniMax). `true` = keep developer.
+ * Per-protocol overrides live under `compatByProtocol[sourceFormat]` when `sourceFormat` matches.
  */
 export function getModelPreserveOpenAIDeveloperRole(
   providerId: string,
-  modelId: string
+  modelId: string,
+  sourceFormat?: string | null
 ): boolean | undefined {
   const m = getCustomModelRow(providerId, modelId);
+  const protocol = sourceFormat && isCompatProtocolKey(sourceFormat) ? sourceFormat : null;
+
   if (m) {
+    if (protocol) {
+      const pc = (m.compatByProtocol as CompatByProtocolMap | undefined)?.[protocol];
+      if (pc && Object.prototype.hasOwnProperty.call(pc, "preserveOpenAIDeveloperRole")) {
+        return Boolean(pc.preserveOpenAIDeveloperRole);
+      }
+    }
     if (Object.prototype.hasOwnProperty.call(m, "preserveOpenAIDeveloperRole")) {
       return Boolean(m.preserveOpenAIDeveloperRole);
     }
     return undefined;
   }
   const co = readCompatList(providerId).find((e) => e.id === modelId);
+  if (protocol && co?.compatByProtocol?.[protocol]) {
+    const pc = co.compatByProtocol[protocol]!;
+    if (Object.prototype.hasOwnProperty.call(pc, "preserveOpenAIDeveloperRole")) {
+      return Boolean(pc.preserveOpenAIDeveloperRole);
+    }
+  }
   if (co && Object.prototype.hasOwnProperty.call(co, "preserveOpenAIDeveloperRole")) {
     return Boolean(co.preserveOpenAIDeveloperRole);
   }

src/lib/db/providers.ts

@@ -513,3 +513,98 @@ export async function deleteProviderNode(id: string) {
   backupDbFile("pre-write");
   return rowToCamel(existing);
 }
+
+// ──────────────── T05: Rate-Limit DB Persistence ──────────────────────────
+// Allows rate-limit state to survive token refresh without being accidentally
+// cleared. DB column rate_limited_until already exists in schema.
+// Ref: sub2api PR #1218 (fix(openai): prevent rescheduling rate-limited accounts)
+
+/**
+ * T05: Persist when a connection is rate-limited, directly in DB.
+ * This survives token refresh — OAuth flows must NOT override this field.
+ *
+ * @param connectionId - The provider_connections.id
+ * @param until - Epoch ms when the rate limit expires (null to clear)
+ */
+export function setConnectionRateLimitUntil(connectionId: string, until: number | null): void {
+  const db = getDbInstance() as unknown as DbLike;
+  db.prepare(
+    "UPDATE provider_connections SET rate_limited_until = ?, updated_at = ? WHERE id = ?"
+  ).run(until, new Date().toISOString(), connectionId);
+  invalidateDbCache("connections");
+}
+
+/**
+ * T05: Check if a connection is currently rate-limited (DB-backed).
+ * Use this before account selection to skip transiently rate-limited accounts.
+ *
+ * @returns true if rate_limited_until is set and in the future
+ */
+export function isConnectionRateLimited(connectionId: string): boolean {
+  const db = getDbInstance() as unknown as DbLike;
+  const row = db
+    .prepare("SELECT rate_limited_until FROM provider_connections WHERE id = ?")
+    .get(connectionId) as { rate_limited_until?: number | null } | undefined;
+  if (!row?.rate_limited_until) return false;
+  return Date.now() < row.rate_limited_until;
+}
+
+/**
+ * T05: Get all connections for a provider that are currently rate-limited.
+ * Returns an array of { id, rateLimitedUntil } for dashboard display.
+ */
+export function getRateLimitedConnections(
+  provider: string
+): Array<{ id: string; rateLimitedUntil: number }> {
+  const db = getDbInstance() as unknown as DbLike;
+  const now = Date.now();
+  const rows = db
+    .prepare(
+      "SELECT id, rate_limited_until FROM provider_connections WHERE provider = ? AND rate_limited_until > ?"
+    )
+    .all(provider, now) as Array<{ id: string; rate_limited_until: number }>;
+  return rows.map((r) => ({ id: r.id, rateLimitedUntil: r.rate_limited_until }));
+}
+
+// ──────────────── T13: Stale Quota Display Fix ─────────────────────────────
+// Codex/Claude quotas display stale cumulative usage after the window resets.
+// By comparing resetAt timestamp to now(), we can show 0 when window has passed.
+// Ref: sub2api PR #1171 (fix: quota display shows stale cumulative usage after reset)
+
+/**
+ * T13: Get effective quota usage, zeroing it out if the window has already reset.
+ *
+ * @param used - Stored usage value (tokens used in the window)
+ * @param resetAt - ISO-8601 string or epoch ms when the window resets, or null
+ * @returns Effective usage: 0 if window expired, original value otherwise
+ */
+export function getEffectiveQuotaUsage(
+  used: number,
+  resetAt: string | number | null | undefined
+): number {
+  if (!resetAt) return used;
+  const resetTime = typeof resetAt === "number" ? resetAt : new Date(resetAt).getTime();
+  if (isNaN(resetTime)) return used;
+  // Window has passed — display should show 0 (pending next snapshot)
+  if (Date.now() >= resetTime) return 0;
+  return used;
+}
+
+/**
+ * T13: Format a reset countdown as a human-readable string: "2h 35m" or "4m 30s".
+ * Returns null if resetAt is in the past or not set.
+ */
+export function formatResetCountdown(resetAt: string | number | null | undefined): string | null {
+  if (!resetAt) return null;
+  const resetTime = typeof resetAt === "number" ? resetAt : new Date(resetAt).getTime();
+  if (isNaN(resetTime)) return null;
+  const diffMs = resetTime - Date.now();
+  if (diffMs <= 0) return null;
+  const totalSeconds = Math.floor(diffMs / 1000);
+  const hours = Math.floor(totalSeconds / 3600);
+  const minutes = Math.floor((totalSeconds % 3600) / 60);
+  const seconds = totalSeconds % 60;
+  if (hours > 0) return `${hours}h ${minutes}m`;
+  if (minutes > 0) return `${minutes}m ${seconds}s`;
+  return `${seconds}s`;
+}

src/lib/db/registeredKeys.ts

@@ -0,0 +1,531 @@
+/**
+ * db/registeredKeys.ts — Registered Keys Provisioning (#464)
+ *
+ * Handles:
+ *   - Issuing registered keys with idempotency
+ *   - Per-provider and per-account quota enforcement
+ *   - Key revocation
+ *   - Quota status queries for rate-limiting decisions
+ */
+
+import { createHash, randomBytes } from "crypto";
+import { v4 as uuidv4 } from "uuid";
+import { getDbInstance, rowToCamel } from "./core";
+
+// ─── Types ───────────────────────────────────────────────────────────────────
+
+export interface RegisteredKey {
+  id: string;
+  keyPrefix: string;
+  name: string;
+  provider: string;
+  accountId: string;
+  isActive: boolean;
+  revokedAt: string | null;
+  expiresAt: string | null;
+  idempotencyKey: string | null;
+  dailyBudget: number | null;
+  hourlyBudget: number | null;
+  dailyUsed: number;
+  hourlyUsed: number;
+  createdAt: string;
+  updatedAt: string;
+}
+
+export interface RegisteredKeyWithSecret extends RegisteredKey {
+  /** Raw key material — only returned once on creation */
+  rawKey: string;
+}
+
+export interface ProviderKeyLimit {
+  provider: string;
+  maxActiveKeys: number | null;
+  dailyIssueLimit: number | null;
+  hourlyIssueLimit: number | null;
+  dailyIssued: number;
+  hourlyIssued: number;
+  updatedAt: string;
+}
+
+export interface AccountKeyLimit {
+  accountId: string;
+  maxActiveKeys: number | null;
+  dailyIssueLimit: number | null;
+  hourlyIssueLimit: number | null;
+  dailyIssued: number;
+  hourlyIssued: number;
+  updatedAt: string;
+}
+
+export interface QuotaCheckResult {
+  allowed: boolean;
+  errorCode?: string;
+  errorMessage?: string;
+  provider?: string;
+  accountId?: string;
+  providerActiveKeys?: number;
+  accountActiveKeys?: number;
+}
+
+export interface IssueKeyParams {
+  name: string;
+  provider?: string;
+  accountId?: string;
+  idempotencyKey?: string;
+  expiresAt?: string;
+  dailyBudget?: number;
+  hourlyBudget?: number;
+}
+
+// ─── Helpers ─────────────────────────────────────────────────────────────────
+
+function nowDay(): string {
+  return new Date().toISOString().slice(0, 10); // YYYY-MM-DD
+}
+
+function nowHour(): string {
+  return new Date().toISOString().slice(0, 13); // YYYY-MM-DDTHH
+}
+
+function hashKey(raw: string): string {
+  return createHash("sha256").update(raw).digest("hex");
+}
+
+function generateRawKey(): string {
+  // ork_ prefix so users can easily identify these keys
+  return "ork_" + randomBytes(24).toString("base64url");
+}
+
+/** Reset window counters if the tracking period has changed. */
+function maybeResetWindow(
+  db: ReturnType<typeof getDbInstance>,
+  table: string,
+  idField: string,
+  idValue: string
+): void {
+  const today = nowDay();
+  const hour = nowHour();
+
+  db.prepare(
+    `
+    UPDATE ${table}
+    SET daily_issued = CASE WHEN last_reset_day <> ? THEN 0 ELSE daily_issued END,
+        hourly_issued = CASE WHEN last_reset_hour <> ? THEN 0 ELSE hourly_issued END,
+        last_reset_day = ?,
+        last_reset_hour = ?
+    WHERE ${idField} = ?
+  `
+  ).run(today, hour, today, hour, idValue);
+}
+
+// ─── Public API ──────────────────────────────────────────────────────────────
+
+/**
+ * Check if a new registered key can be issued for the given provider/account.
+ * Returns { allowed: true } or { allowed: false, errorCode, errorMessage }.
+ */
+export function checkQuota(provider = "", accountId = ""): QuotaCheckResult {
+  const db = getDbInstance();
+  const today = nowDay();
+  const hour = nowHour();
+
+  // ── provider-level check ──
+  if (provider) {
+    maybeResetWindow(db, "provider_key_limits", "provider", provider);
+
+    const limits = db
+      .prepare("SELECT * FROM provider_key_limits WHERE provider = ?")
+      .get(provider) as ProviderKeyLimitRow | undefined;
+
+    if (limits) {
+      if (limits.hourly_issue_limit !== null && limits.hourly_issued >= limits.hourly_issue_limit) {
+        return {
+          allowed: false,
+          errorCode: "PROVIDER_QUOTA_EXCEEDED",
+          errorMessage: `Hourly issue limit (${limits.hourly_issue_limit}) reached for provider '${provider}'`,
+          provider,
+        };
+      }
+      if (limits.daily_issue_limit !== null && limits.daily_issued >= limits.daily_issue_limit) {
+        return {
+          allowed: false,
+          errorCode: "PROVIDER_QUOTA_EXCEEDED",
+          errorMessage: `Daily issue limit (${limits.daily_issue_limit}) reached for provider '${provider}'`,
+          provider,
+        };
+      }
+      if (limits.max_active_keys !== null) {
+        const { activeCount } = db
+          .prepare(
+            "SELECT COUNT(*) as activeCount FROM registered_keys WHERE provider = ? AND is_active = 1"
+          )
+          .get(provider) as { activeCount: number };
+        if (activeCount >= limits.max_active_keys) {
+          return {
+            allowed: false,
+            errorCode: "MAX_ACTIVE_KEYS_EXCEEDED",
+            errorMessage: `Max active keys (${limits.max_active_keys}) reached for provider '${provider}'`,
+            provider,
+            providerActiveKeys: activeCount,
+          };
+        }
+      }
+    }
+  }
+
+  // ── account-level check ──
+  if (accountId) {
+    maybeResetWindow(db, "account_key_limits", "account_id", accountId);
+
+    const limits = db
+      .prepare("SELECT * FROM account_key_limits WHERE account_id = ?")
+      .get(accountId) as AccountKeyLimitRow | undefined;
+
+    if (limits) {
+      if (limits.hourly_issue_limit !== null && limits.hourly_issued >= limits.hourly_issue_limit) {
+        return {
+          allowed: false,
+          errorCode: "ACCOUNT_QUOTA_EXCEEDED",
+          errorMessage: `Hourly issue limit (${limits.hourly_issue_limit}) reached for account '${accountId}'`,
+          accountId,
+        };
+      }
+      if (limits.daily_issue_limit !== null && limits.daily_issued >= limits.daily_issue_limit) {
+        return {
+          allowed: false,
+          errorCode: "ACCOUNT_QUOTA_EXCEEDED",
+          errorMessage: `Daily issue limit (${limits.daily_issue_limit}) reached for account '${accountId}'`,
+          accountId,
+        };
+      }
+      if (limits.max_active_keys !== null) {
+        const { activeCount } = db
+          .prepare(
+            "SELECT COUNT(*) as activeCount FROM registered_keys WHERE account_id = ? AND is_active = 1"
+          )
+          .get(accountId) as { activeCount: number };
+        if (activeCount >= limits.max_active_keys) {
+          return {
+            allowed: false,
+            errorCode: "MAX_ACTIVE_KEYS_EXCEEDED",
+            errorMessage: `Max active keys (${limits.max_active_keys}) reached for account '${accountId}'`,
+            accountId,
+            accountActiveKeys: activeCount,
+          };
+        }
+      }
+    }
+  }
+
+  return { allowed: true };
+}
+
+/**
+ * Issue a new registered key.
+ * Returns the key with rawKey (only on creation) or null if idempotency_key already exists.
+ */
+export function issueRegisteredKey(
+  params: IssueKeyParams
+): RegisteredKeyWithSecret | { idempotencyConflict: true; existing: RegisteredKey } {
+  const db = getDbInstance();
+  const {
+    name,
+    provider = "",
+    accountId = "",
+    idempotencyKey,
+    expiresAt,
+    dailyBudget,
+    hourlyBudget,
+  } = params;
+
+  // ── idempotency check ──
+  if (idempotencyKey) {
+    const existing = db
+      .prepare("SELECT * FROM registered_keys WHERE idempotency_key = ?")
+      .get(idempotencyKey) as RegisteredKeyRow | undefined;
+    if (existing) {
+      return {
+        idempotencyConflict: true,
+        existing: rowToCamel(existing) as unknown as RegisteredKey,
+      };
+    }
+  }
+
+  const rawKey = generateRawKey();
+  const id = uuidv4();
+  const keyHash = hashKey(rawKey);
+  const keyPrefix = rawKey.slice(0, 12); // "ork_" + 8 chars
+
+  db.prepare(
+    `
+    INSERT INTO registered_keys
+      (id, key, key_prefix, name, provider, account_id, idempotency_key, expires_at, daily_budget, hourly_budget, last_reset_day, last_reset_hour)
+    VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+  `
+  ).run(
+    id,
+    keyHash,
+    keyPrefix,
+    name,
+    provider,
+    accountId,
+    idempotencyKey ?? null,
+    expiresAt ?? null,
+    dailyBudget ?? null,
+    hourlyBudget ?? null,
+    nowDay(),
+    nowHour()
+  );
+
+  // Increment provider/account issuance counters
+  if (provider) {
+    maybeResetWindow(db, "provider_key_limits", "provider", provider);
+    db.prepare(
+      `
+      INSERT INTO provider_key_limits (provider, daily_issued, hourly_issued, last_reset_day, last_reset_hour)
+      VALUES (?, 1, 1, ?, ?)
+      ON CONFLICT(provider) DO UPDATE SET
+        daily_issued = daily_issued + 1,
+        hourly_issued = hourly_issued + 1,
+        updated_at = datetime('now')
+    `
+    ).run(provider, nowDay(), nowHour());
+  }
+  if (accountId) {
+    maybeResetWindow(db, "account_key_limits", "account_id", accountId);
+    db.prepare(
+      `
+      INSERT INTO account_key_limits (account_id, daily_issued, hourly_issued, last_reset_day, last_reset_hour)
+      VALUES (?, 1, 1, ?, ?)
+      ON CONFLICT(account_id) DO UPDATE SET
+        daily_issued = daily_issued + 1,
+        hourly_issued = hourly_issued + 1,
+        updated_at = datetime('now')
+    `
+    ).run(accountId, nowDay(), nowHour());
+  }
+
+  const created = db
+    .prepare("SELECT * FROM registered_keys WHERE id = ?")
+    .get(id) as RegisteredKeyRow;
+  return { ...(rowToCamel(created) as unknown as RegisteredKey), rawKey };
+}
+
+/**
+ * Get a registered key by ID (without the raw key — only prefix is returned).
+ */
+export function getRegisteredKey(id: string): RegisteredKey | null {
+  const db = getDbInstance();
+  const row = db.prepare("SELECT * FROM registered_keys WHERE id = ?").get(id) as
+    | RegisteredKeyRow
+    | undefined;
+  return row ? (rowToCamel(row) as unknown as RegisteredKey) : null;
+}
+
+/**
+ * List all registered keys (optionally filtered by provider/accountId).
+ */
+export function listRegisteredKeys(
+  opts: { provider?: string; accountId?: string } = {}
+): RegisteredKey[] {
+  const db = getDbInstance();
+  let sql = "SELECT * FROM registered_keys WHERE 1=1";
+  const args: string[] = [];
+  if (opts.provider) {
+    sql += " AND provider = ?";
+    args.push(opts.provider);
+  }
+  if (opts.accountId) {
+    sql += " AND account_id = ?";
+    args.push(opts.accountId);
+  }
+  sql += " ORDER BY created_at DESC LIMIT 500";
+  const rows = db.prepare(sql).all(...args) as RegisteredKeyRow[];
+  return rows.map((r) => rowToCamel(r) as unknown as RegisteredKey);
+}
+
+/**
+ * Revoke a registered key by ID.
+ */
+export function revokeRegisteredKey(id: string): boolean {
+  const db = getDbInstance();
+  const result = db
+    .prepare(
+      `
+    UPDATE registered_keys
+    SET is_active = 0, revoked_at = datetime('now'), updated_at = datetime('now')
+    WHERE id = ? AND is_active = 1
+  `
+    )
+    .run(id);
+  return result.changes > 0;
+}
+
+/**
+ * Validate a raw registered key against stored hashes.
+ * Returns the key metadata if valid, null otherwise.
+ */
+export function validateRegisteredKey(rawKey: string): RegisteredKey | null {
+  const db = getDbInstance();
+  const hash = hashKey(rawKey);
+  const row = db
+    .prepare(
+      `
+    SELECT * FROM registered_keys
+    WHERE key = ? AND is_active = 1
+      AND (expires_at IS NULL OR expires_at > datetime('now'))
+  `
+    )
+    .get(hash) as RegisteredKeyRow | undefined;
+  if (!row) return null;
+
+  // Auto-reset budget windows if needed
+  const today = nowDay();
+  const hour = nowHour();
+  if (row.last_reset_day !== today || row.last_reset_hour !== hour) {
+    db.prepare(
+      `
+      UPDATE registered_keys
+      SET daily_used = CASE WHEN last_reset_day <> ? THEN 0 ELSE daily_used END,
+          hourly_used = CASE WHEN last_reset_hour <> ? THEN 0 ELSE hourly_used END,
+          last_reset_day = ?, last_reset_hour = ?
+      WHERE id = ?
+    `
+    ).run(today, hour, today, hour, row.id);
+  }
+
+  // Budget check
+  if (row.daily_budget !== null && row.daily_used >= row.daily_budget) return null;
+  if (row.hourly_budget !== null && row.hourly_used >= row.hourly_budget) return null;
+
+  return rowToCamel(row) as unknown as RegisteredKey;
+}
+
+/**
+ * Increment usage counters for a registered key (called by request pipeline).
+ */
+export function incrementRegisteredKeyUsage(id: string): void {
+  const db = getDbInstance();
+  db.prepare(
+    `
+    UPDATE registered_keys
+    SET daily_used = daily_used + 1, hourly_used = hourly_used + 1, updated_at = datetime('now')
+    WHERE id = ?
+  `
+  ).run(id);
+}
+
+// ─── Provider / Account Limit Management ──────────────────────────────────────
+
+export function setProviderKeyLimit(
+  provider: string,
+  limits: Partial<Omit<ProviderKeyLimit, "provider" | "dailyIssued" | "hourlyIssued" | "updatedAt">>
+): void {
+  const db = getDbInstance();
+  db.prepare(
+    `
+    INSERT INTO provider_key_limits (provider, max_active_keys, daily_issue_limit, hourly_issue_limit, last_reset_day, last_reset_hour)
+    VALUES (?, ?, ?, ?, ?, ?)
+    ON CONFLICT(provider) DO UPDATE SET
+      max_active_keys = excluded.max_active_keys,
+      daily_issue_limit = excluded.daily_issue_limit,
+      hourly_issue_limit = excluded.hourly_issue_limit,
+      updated_at = datetime('now')
+  `
+  ).run(
+    provider,
+    limits.maxActiveKeys ?? null,
+    limits.dailyIssueLimit ?? null,
+    limits.hourlyIssueLimit ?? null,
+    nowDay(),
+    nowHour()
+  );
+}
+
+export function setAccountKeyLimit(
+  accountId: string,
+  limits: Partial<Omit<AccountKeyLimit, "accountId" | "dailyIssued" | "hourlyIssued" | "updatedAt">>
+): void {
+  const db = getDbInstance();
+  db.prepare(
+    `
+    INSERT INTO account_key_limits (account_id, max_active_keys, daily_issue_limit, hourly_issue_limit, last_reset_day, last_reset_hour)
+    VALUES (?, ?, ?, ?, ?, ?)
+    ON CONFLICT(account_id) DO UPDATE SET
+      max_active_keys = excluded.max_active_keys,
+      daily_issue_limit = excluded.daily_issue_limit,
+      hourly_issue_limit = excluded.hourly_issue_limit,
+      updated_at = datetime('now')
+  `
+  ).run(
+    accountId,
+    limits.maxActiveKeys ?? null,
+    limits.dailyIssueLimit ?? null,
+    limits.hourlyIssueLimit ?? null,
+    nowDay(),
+    nowHour()
+  );
+}
+
+export function getProviderKeyLimit(provider: string): ProviderKeyLimit | null {
+  const db = getDbInstance();
+  const row = db.prepare("SELECT * FROM provider_key_limits WHERE provider = ?").get(provider) as
+    | ProviderKeyLimitRow
+    | undefined;
+  return row ? (rowToCamel(row) as unknown as ProviderKeyLimit) : null;
+}
+
+export function getAccountKeyLimit(accountId: string): AccountKeyLimit | null {
+  const db = getDbInstance();
+  const row = db.prepare("SELECT * FROM account_key_limits WHERE account_id = ?").get(accountId) as
+    | AccountKeyLimitRow
+    | undefined;
+  return row ? (rowToCamel(row) as unknown as AccountKeyLimit) : null;
+}
+
+// ─── Internal types (raw DB rows) ─────────────────────────────────────────────
+
+interface RegisteredKeyRow {
+  id: string;
+  key: string;
+  key_prefix: string;
+  name: string;
+  provider: string;
+  account_id: string;
+  is_active: number;
+  revoked_at: string | null;
+  expires_at: string | null;
+  idempotency_key: string | null;
+  daily_budget: number | null;
+  hourly_budget: number | null;
+  daily_used: number;
+  hourly_used: number;
+  last_reset_day: string;
+  last_reset_hour: string;
+  created_at: string;
+  updated_at: string;
+}
+
+interface ProviderKeyLimitRow {
+  provider: string;
+  max_active_keys: number | null;
+  daily_issue_limit: number | null;
+  hourly_issue_limit: number | null;
+  daily_issued: number;
+  hourly_issued: number;
+  last_reset_day: string;
+  last_reset_hour: string;
+  updated_at: string;
+}
+
+interface AccountKeyLimitRow {
+  account_id: string;
+  max_active_keys: number | null;
+  daily_issue_limit: number | null;
+  hourly_issue_limit: number | null;
+  daily_issued: number;
+  hourly_issued: number;
+  last_reset_day: string;
+  last_reset_hour: string;
+  updated_at: string;
+}

src/lib/ipUtils.ts

@@ -0,0 +1,56 @@
+import { isIP } from "node:net";
+
+/**
+ * T07: Extract the real client IP from X-Forwarded-For header.
+ * Skips invalid entries like "unknown" or empty strings.
+ * Falls back to remoteAddress if no valid IP found.
+ * Ref: sub2api PR #1135
+ *
+ * @param xForwardedFor - Value of the X-Forwarded-For header (may be CSV)
+ * @param remoteAddress - Fallback from the raw socket (req.socket.remoteAddress)
+ * @returns The first valid IP address found, or "unknown"
+ */
+export function extractClientIp(
+  xForwardedFor: string | null | undefined,
+  remoteAddress: string | undefined
+): string {
+  if (xForwardedFor) {
+    const entries = xForwardedFor.split(",");
+    for (const entry of entries) {
+      const trimmed = entry.trim();
+      if (trimmed && isIP(trimmed) !== 0) {
+        return trimmed; // First valid IP wins
+      }
+    }
+  }
+  return remoteAddress?.trim() ?? "unknown";
+}
+
+/**
+ * Extract client IP from a Request or NextRequest object.
+ * Checks X-Forwarded-For, X-Real-IP, CF-Connecting-IP, then socket.
+ */
+export function getClientIpFromRequest(req: {
+  headers?: Headers | { get?: (n: string) => string | null };
+  socket?: { remoteAddress?: string };
+  ip?: string;
+}): string {
+  // Helper to get header value from either Headers object or plain object
+  const getHeader = (name: string): string | null => {
+    if (!req.headers) return null;
+    if (typeof (req.headers as Headers).get === "function") {
+      return (req.headers as Headers).get(name);
+    }
+    return null;
+  };
+
+  // Priority: CF-Connecting-IP (Cloudflare) > X-Forwarded-For > X-Real-IP > socket
+  const cfIp = getHeader("cf-connecting-ip");
+  if (cfIp && isIP(cfIp.trim()) !== 0) return cfIp.trim();
+
+  const xff = getHeader("x-forwarded-for");
+  const realIp = getHeader("x-real-ip");
+  const remoteAddress = req.ip ?? req.socket?.remoteAddress;
+
+  return extractClientIp(xff ?? realIp, remoteAddress);
+}

src/lib/localDb.ts

@@ -23,6 +23,15 @@ export {
   createProviderNode,
   updateProviderNode,
   deleteProviderNode,
+
+  // T05: Rate-limit DB persistence (survives token refresh)
+  setConnectionRateLimitUntil,
+  isConnectionRateLimited,
+  getRateLimitedConnections,
+
+  // T13: Stale quota display fix (zero out usage after window resets)
+  getEffectiveQuotaUsage,
+  formatResetCountdown,
 } from "./db/providers";
 
 export {
@@ -48,6 +57,8 @@ export {
   getModelPreserveOpenAIDeveloperRole,
 } from "./db/models";
 
+export type { ModelCompatPerProtocol, ModelCompatPatch } from "./db/models";
+
 export {
   // Combos
   getCombos,
@@ -136,3 +147,27 @@ export {
   getCachedProviderConnections,
   invalidateDbCache,
 } from "./db/readCache";
+
+export {
+  // Registered Keys Provisioning (#464)
+  issueRegisteredKey,
+  getRegisteredKey,
+  listRegisteredKeys,
+  revokeRegisteredKey,
+  validateRegisteredKey,
+  incrementRegisteredKeyUsage,
+  checkQuota,
+  setProviderKeyLimit,
+  setAccountKeyLimit,
+  getProviderKeyLimit,
+  getAccountKeyLimit,
+} from "./db/registeredKeys";
+
+export type {
+  RegisteredKey,
+  RegisteredKeyWithSecret,
+  ProviderKeyLimit,
+  AccountKeyLimit,
+  QuotaCheckResult,
+  IssueKeyParams,
+} from "./db/registeredKeys";

src/lib/oauth/providers/gemini.ts

@@ -25,6 +25,18 @@ export const gemini = {
 
     if (config.clientSecret) {
       bodyParams.client_secret = config.clientSecret;
+    } else {
+      // (#537) Google's OAuth2 token endpoint always requires client_secret for
+      // non-PKCE flows. Without it we get a cryptic "client_secret is missing" error.
+      // This typically happens in self-hosted / Docker deployments where
+      // GEMINI_OAUTH_CLIENT_SECRET is not set in the container environment.
+      throw new Error(
+        "Gemini CLI OAuth requires GEMINI_OAUTH_CLIENT_SECRET to be set.\n" +
+          "In Docker: add 'GEMINI_OAUTH_CLIENT_SECRET=<your-secret>' to your docker-compose.yml env.\n" +
+          "In npm: add it to ~/.omniroute/.env\n" +
+          "Obtain the client secret from https://console.cloud.google.com/apis/credentials\n" +
+          "for the same OAuth 2.0 Client ID configured as GEMINI_OAUTH_CLIENT_ID."
+      );
     }
 
     const response = await fetch(config.tokenUrl, {

src/lib/providers/validation.ts

@@ -610,7 +610,12 @@ export async function validateProviderApiKey({ provider, apiKey, providerSpecifi
   }
 
   const modelId = entry.models?.[0]?.id || null;
-  const baseUrl = resolveBaseUrl(entry, providerSpecificData);
+  // (#532) Use testKeyBaseUrl if defined — some providers validate keys on a different endpoint
+  // than where requests are sent (e.g. opencode-go validates on zen/v1, not zen/go/v1)
+  const validationEntry = entry.testKeyBaseUrl
+    ? { ...entry, baseUrl: entry.testKeyBaseUrl }
+    : entry;
+  const baseUrl = resolveBaseUrl(validationEntry, providerSpecificData);
 
   try {
     if (OPENAI_LIKE_FORMATS.has(entry.format)) {

src/lib/proxyHealth.ts

@@ -0,0 +1,140 @@
+/**
+ * T14: Proxy Fast-Fail — TCP health check with in-memory cache.
+ *
+ * When a configured HTTP/SOCKS5 proxy is unreachable, every request
+ * through OmniRoute used to wait for the full PROXY_TIMEOUT_MS (30s)
+ * before failing. This module detects dead proxies in <2s via a quick
+ * TCP connection check, caching the result to avoid overhead per request.
+ *
+ * Ref: sub2api PR #1167 (fix: proxy-fast-fail)
+ */
+
+import { createConnection } from "node:net";
+
+// Configurable via env vars
+const FAST_FAIL_TIMEOUT_MS = parseInt(process.env.PROXY_FAST_FAIL_TIMEOUT_MS ?? "2000", 10);
+const HEALTH_CACHE_TTL_MS = parseInt(process.env.PROXY_HEALTH_CACHE_TTL_MS ?? "30000", 10);
+
+interface ProxyHealthEntry {
+  healthy: boolean;
+  checkedAt: number;
+  ttlMs: number;
+}
+
+// In-memory cache: proxyUrl → health entry
+const proxyHealthCache = new Map<string, ProxyHealthEntry>();
+
+/**
+ * T14: Perform a fast TCP check to see if a proxy host:port is reachable.
+ * Results are cached for `cacheTtlMs` (default 30s) to avoid checking every request.
+ *
+ * @param proxyUrl - Full proxy URL, e.g. http://user:pass@1.2.3.4:8080
+ * @param timeoutMs - TCP connection timeout (default 2000ms)
+ * @param cacheTtlMs - How long to cache the health result (default 30000ms)
+ * @returns true if proxy TCP port is open, false otherwise
+ */
+export async function isProxyReachable(
+  proxyUrl: string,
+  timeoutMs = FAST_FAIL_TIMEOUT_MS,
+  cacheTtlMs = HEALTH_CACHE_TTL_MS
+): Promise<boolean> {
+  const cached = proxyHealthCache.get(proxyUrl);
+  if (cached && Date.now() - cached.checkedAt < cached.ttlMs) {
+    return cached.healthy;
+  }
+
+  let url: URL;
+  try {
+    url = new URL(proxyUrl);
+  } catch {
+    // Malformed URL — treat as unreachable
+    proxyHealthCache.set(proxyUrl, {
+      healthy: false,
+      checkedAt: Date.now(),
+      ttlMs: cacheTtlMs,
+    });
+    return false;
+  }
+
+  const host = url.hostname;
+  const port = parseInt(url.port || defaultPortForScheme(url.protocol), 10);
+
+  if (!host || isNaN(port)) {
+    proxyHealthCache.set(proxyUrl, {
+      healthy: false,
+      checkedAt: Date.now(),
+      ttlMs: cacheTtlMs,
+    });
+    return false;
+  }
+
+  const healthy = await tcpCheck(host, port, timeoutMs);
+  proxyHealthCache.set(proxyUrl, { healthy, checkedAt: Date.now(), ttlMs: cacheTtlMs });
+  return healthy;
+}
+
+/**
+ * Get the cached health status of a proxy without re-checking.
+ * Returns null if there is no cached entry.
+ */
+export function getCachedProxyHealth(proxyUrl: string): boolean | null {
+  const cached = proxyHealthCache.get(proxyUrl);
+  if (!cached) return null;
+  if (Date.now() - cached.checkedAt >= cached.ttlMs) return null; // stale
+  return cached.healthy;
+}
+
+/**
+ * Invalidate the cached health for a proxy URL (force re-check on next call).
+ */
+export function invalidateProxyHealth(proxyUrl: string): void {
+  proxyHealthCache.delete(proxyUrl);
+}
+
+/**
+ * Get all currently cached proxy health entries (for dashboard display).
+ */
+export function getAllProxyHealthStatuses(): Array<{
+  proxyUrl: string;
+  healthy: boolean;
+  checkedAt: number;
+  stale: boolean;
+}> {
+  const now = Date.now();
+  return [...proxyHealthCache.entries()].map(([proxyUrl, entry]) => ({
+    proxyUrl,
+    healthy: entry.healthy,
+    checkedAt: entry.checkedAt,
+    stale: now - entry.checkedAt >= entry.ttlMs,
+  }));
+}
+
+// ─── Internals ────────────────────────────────────────────────────────────────
+
+function defaultPortForScheme(protocol: string): string {
+  switch (protocol.replace(":", "").toLowerCase()) {
+    case "https":
+      return "443";
+    case "socks5":
+    case "socks5h":
+      return "1080";
+    case "http":
+    default:
+      return "8080";
+  }
+}
+
+function tcpCheck(host: string, port: number, timeoutMs: number): Promise<boolean> {
+  return new Promise<boolean>((resolve) => {
+    const socket = createConnection({ host, port }, () => {
+      socket.destroy();
+      resolve(true);
+    });
+    socket.setTimeout(timeoutMs);
+    socket.on("error", () => resolve(false));
+    socket.on("timeout", () => {
+      socket.destroy();
+      resolve(false);
+    });
+  });
+}

src/lib/usage/callLogs.ts

@@ -180,6 +180,7 @@ export async function saveCallLog(entry: any) {
       path: entry.path || "/v1/chat/completions",
       status: entry.status || 0,
       model: entry.model || "-",
+      requestedModel: entry.requestedModel || null, // T01: model the client asked for
       provider: entry.provider || "-",
       account,
       connectionId: entry.connectionId || null,
@@ -205,10 +206,10 @@ export async function saveCallLog(entry: any) {
     const db = getDbInstance();
     db.prepare(
       `
-      INSERT INTO call_logs (id, timestamp, method, path, status, model, provider,
+      INSERT INTO call_logs (id, timestamp, method, path, status, model, requested_model, provider,
         account, connection_id, duration, tokens_in, tokens_out, request_type, source_format, target_format,
         api_key_id, api_key_name, combo_name, request_body, response_body, error)
-      VALUES (@id, @timestamp, @method, @path, @status, @model, @provider,
+      VALUES (@id, @timestamp, @method, @path, @status, @model, @requestedModel, @provider,
         @account, @connectionId, @duration, @tokensIn, @tokensOut, @requestType, @sourceFormat, @targetFormat,
         @apiKeyId, @apiKeyName, @comboName, @requestBody, @responseBody, @error)
     `
@@ -374,6 +375,7 @@ export async function getCallLogs(filter: any = {}) {
       path: toStringOrNull(l.path),
       status: toNumber(l.status),
       model: toStringOrNull(l.model),
+      requestedModel: toStringOrNull(l.requested_model), // T01: original model from client
       provider: toStringOrNull(l.provider),
       account: toStringOrNull(l.account),
       duration: toNumber(l.duration),