Repository navigation
Add support for latest AI models (Claude 4.6, GPT-5.3, Gemini 3.1, Qwen3.5) - #747
Conversation
Add gpt-5.3-chat-latest (released March 3, 2026) to the OpenAI direct provider and OpenRouter paid provider, with vision capability entries. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
Add gemini-3.1-flash-lite-preview (released Mar 3, 2026) with 1M context window to Google direct and OpenRouter providers, with vision capability. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
…iders Add gemini-3.1-pro-preview-customtools (released Feb 25, 2026) with 1M context window to Google direct and OpenRouter providers, with vision. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
Add openai/gpt-5.3-codex (released Feb 24, 2026, 400K context) to the OpenRouter paid provider with vision capability entry. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
Add gemini-3.1-pro-preview (released Feb 19, 2026) with 1M context window to Google direct and OpenRouter providers, with vision capability. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
Add claude-sonnet-4-6-20260217 (released Feb 17, 2026) with 1M context to Anthropic direct and OpenRouter providers, with vision capability. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
Add four Qwen3.5 native vision-language models (released Feb 25, 2026): - qwen3.5-flash (1M context) - qwen3.5-122b-a10b (262K context) - qwen3.5-35b-a3b (262K context) - qwen3.5-27b (262K context) All are vision-capable models with hybrid linear attention + MoE architecture. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
Add two more Qwen3.5 native vision-language models (released Feb 16, 2026): - qwen3.5-397b-a17b (262K context, flagship model) - qwen3.5-plus-2026-02-15 (1M context) https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
Add qwen/qwen3-max-thinking (released Feb 9, 2026, 262K context), the flagship reasoning model in the Qwen3 series. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
|
Warning Rate limit exceeded
⌛ How to resolve this issue?After the wait time has elapsed, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout. Please see our FAQ for further information. ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (4)
📝 WalkthroughWalkthroughAdded many new model entries (GPT‑5.4/5.3, Claude Sonnet 4.6, Gemini 3.1 variants, Qwen3.5 family, Mercury 2) across AI provider registries and extended the vision-capable models registry; also adjusted IntegrationAuditLogPage stats/export logic and its test. Changes
Sequence Diagram(s)(omitted) Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 2 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: bd6c37d717
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| // Qwen3.5 Vision-Language Models | ||
| 'qwen/qwen3.5-397b-a17b': true, | ||
| 'qwen/qwen3.5-plus-2026-02-15': true, | ||
| 'qwen/qwen3.5-flash': true, | ||
| 'qwen/qwen3.5-122b-a10b': true, |
There was a problem hiding this comment.
This Qwen vision-model block omits qwen/qwen3-max-thinking, even though that model was newly added in ai-providers-config.ts alongside the other Qwen3.5 entries; because hasVisionCapability() does not have a Qwen heuristic fallback, this ID will be treated as non-vision at runtime and image input/features will be disabled for users selecting it.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Fixed in 54ac5f9 — added qwen3-max-thinking to the vision model registry. With the prefix-stripping refactor, both qwen/qwen3-max-thinking and bare qwen3-max-thinking are now resolved correctly.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@apps/web/src/lib/ai/core/vision-models.ts`:
- Around line 99-106: VISION_CAPABLE_MODELS is missing the
'qwen/qwen3-max-thinking' entry, causing hasVisionCapability fallback to
mis-classify that model as non-vision-capable; add 'qwen/qwen3-max-thinking':
true to the VISION_CAPABLE_MODELS map/constant (the same structure that contains
other qwen/* entries) so the model is recognized by hasVisionCapability and
image requests are not blocked.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 85457d9e-cb61-44e8-9ece-d67f66b3f343
📒 Files selected for processing (2)
apps/web/src/lib/ai/core/ai-providers-config.tsapps/web/src/lib/ai/core/vision-models.ts
- GPT-5.4: OpenAI's latest frontier model, 1M+ context, vision support - GPT-5.4 Pro: Enhanced reasoning variant for complex tasks, vision support - Mercury 2: Inception's reasoning diffusion LLM, 128K context, >1000 tok/s All released March 4-5, 2026. https://claude.ai/code/session_013JAWABx5QQCSth4ApeU6dH
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
apps/web/src/lib/ai/core/ai-providers-config.ts (1)
63-76:⚠️ Potential issue | 🟠 MajorThese new third-party models still bypass the current subscription gate.
The provided
apps/web/src/app/api/ai/chat/route.tssnippet shows chat requests only consultingrequiresProSubscription(...), and the accompanying context says that helper returnsfalsefor non-PageSpace providers. Adding more Claude/GPT/Gemini/Qwen/Mercury entries here therefore expands the set of premium models users can hit without any entitlement check.Also applies to: 113-119, 133-135, 152-152, 216-218, 239-240, 288-288
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/web/src/lib/ai/core/ai-providers-config.ts` around lines 63 - 76, Third-party model entries added to ai-providers-config.ts are currently bypassing the subscription gate because requiresProSubscription(...) (used in apps/web/src/app/api/ai/chat/route.ts) returns false for non-PageSpace providers; fix by either adding an explicit premium flag to the provider configuration in ai-providers-config.ts (e.g., requiresPro: true for entries like 'anthropic/claude-sonnet-4.6', 'openai/gpt-5.4', etc.) and updating the chat route to consult that flag, or modify the requiresProSubscription function to treat non-PageSpace namespaces (anthropic/, openai/, gemini/, qwen/, mercury/, etc.) as requiring pro access; update the chat route to call the new logic (requiresProSubscription) so these new keys are correctly gated.
♻️ Duplicate comments (1)
apps/web/src/lib/ai/core/vision-models.ts (1)
105-111:⚠️ Potential issue | 🟠 MajorAdd
qwen/qwen3-max-thinkingto the vision allowlist.
apps/web/src/lib/ai/core/ai-providers-config.tsnow exposesqwen/qwen3-max-thinkingat Line 119, buthasVisionCapability()has no Qwen fallback. Without a static entry here, that model resolves tofalseand image requests get blocked.Suggested fix
'qwen/qwen3.5-flash': true, 'qwen/qwen3.5-122b-a10b': true, 'qwen/qwen3.5-35b-a3b': true, 'qwen/qwen3.5-27b': true, + 'qwen/qwen3-max-thinking': true,🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/web/src/lib/ai/core/vision-models.ts` around lines 105 - 111, The vision allowlist in apps/web/src/lib/ai/core/vision-models.ts is missing the Qwen fallback entry so hasVisionCapability() will treat qwen/qwen3-max-thinking as unsupported; add the key 'qwen/qwen3-max-thinking': true to the exported allowlist object (alongside other 'qwen/...' entries such as 'qwen/qwen3.5-397b-a17b') so that the model exposed in ai-providers-config.ts is recognized and image requests are allowed.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@apps/web/src/lib/ai/core/ai-providers-config.ts`:
- Around line 239-240: The AI_PROVIDERS.openai.models registry is missing
GPT-5.4 entries so isValidModel('openai', ...) and the UI selector reject
gpt-5.4 identifiers; update the AI_PROVIDERS.openai.models object to include
mappings for 'gpt-5.4' (friendly name e.g., 'GPT-5.4') and 'gpt-5.4-pro' (e.g.,
'GPT-5.4 Pro') matching the keys used in vision-models.ts so the direct OpenAI
provider recognizes these models.
---
Outside diff comments:
In `@apps/web/src/lib/ai/core/ai-providers-config.ts`:
- Around line 63-76: Third-party model entries added to ai-providers-config.ts
are currently bypassing the subscription gate because
requiresProSubscription(...) (used in apps/web/src/app/api/ai/chat/route.ts)
returns false for non-PageSpace providers; fix by either adding an explicit
premium flag to the provider configuration in ai-providers-config.ts (e.g.,
requiresPro: true for entries like 'anthropic/claude-sonnet-4.6',
'openai/gpt-5.4', etc.) and updating the chat route to consult that flag, or
modify the requiresProSubscription function to treat non-PageSpace namespaces
(anthropic/, openai/, gemini/, qwen/, mercury/, etc.) as requiring pro access;
update the chat route to call the new logic (requiresProSubscription) so these
new keys are correctly gated.
---
Duplicate comments:
In `@apps/web/src/lib/ai/core/vision-models.ts`:
- Around line 105-111: The vision allowlist in
apps/web/src/lib/ai/core/vision-models.ts is missing the Qwen fallback entry so
hasVisionCapability() will treat qwen/qwen3-max-thinking as unsupported; add the
key 'qwen/qwen3-max-thinking': true to the exported allowlist object (alongside
other 'qwen/...' entries such as 'qwen/qwen3.5-397b-a17b') so that the model
exposed in ai-providers-config.ts is recognized and image requests are allowed.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 183d6092-16d9-4835-b65a-7766db9d995f
📒 Files selected for processing (2)
apps/web/src/lib/ai/core/ai-providers-config.tsapps/web/src/lib/ai/core/vision-models.ts
- Fix misleading stats labels: clarify success rate and avg duration are computed from the current page only, not the full dataset - Remove duplicate filter-to-params logic in handleExport by reusing hookParams - Compute stats in a single pass with useMemo instead of separate filter+reduce calls - Remove unnecessary DOM append/remove for CSV download element - Tighten FiltersState.success type from string to union literal - Deduplicate vision-models.ts: replace ~30 provider-prefixed entries with a stripPrefix helper in hasVisionCapability lookup
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
apps/web/src/components/integrations/IntegrationAuditLogPage.tsx (1)
282-287:⚠️ Potential issue | 🟡 MinorType mismatch in Select onValueChange handler.
The
onValueChangecallback receivesvalue: stringfrom the Select component, butfilters.successis typed as'all' | 'true' | 'false'. The assignmentsuccess: valueshould use a type assertion to maintain type safety.Proposed fix
<Select value={filters.success} onValueChange={(value) => { - setFilters((prev) => ({ ...prev, success: value })); + setFilters((prev) => ({ ...prev, success: value as 'all' | 'true' | 'false' })); setCurrentPage(1); }} >🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/web/src/components/integrations/IntegrationAuditLogPage.tsx` around lines 282 - 287, In IntegrationAuditLogPage, the Select onValueChange handler assigns a string to filters.success which is typed as the union 'all' | 'true' | 'false'; fix this by using a TypeScript type assertion when updating filters (cast the incoming value to the 'all' | 'true' | 'false' union) so the setFilters call remains type-safe, and keep the setCurrentPage(1) call as-is.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@apps/web/src/lib/ai/core/vision-models.ts`:
- Around line 74-79: The vision-capable registry is missing the 'grok-4-fast'
entry causing hasVisionCapability to return false for that model; update the
registry object in apps/web/src/lib/ai/core/vision-models.ts (the map that
currently includes 'grok-4', 'grok-2-vision', etc.) to include 'grok-4-fast':
true so hasVisionCapability (the function that checks for both "grok" and
"vision" substrings) will correctly recognize it as vision-capable.
---
Outside diff comments:
In `@apps/web/src/components/integrations/IntegrationAuditLogPage.tsx`:
- Around line 282-287: In IntegrationAuditLogPage, the Select onValueChange
handler assigns a string to filters.success which is typed as the union 'all' |
'true' | 'false'; fix this by using a TypeScript type assertion when updating
filters (cast the incoming value to the 'all' | 'true' | 'false' union) so the
setFilters call remains type-safe, and keep the setCurrentPage(1) call as-is.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 20fa6543-04da-420b-bb69-d27dce83bc77
📒 Files selected for processing (3)
apps/web/src/components/integrations/IntegrationAuditLogPage.tsxapps/web/src/components/integrations/__tests__/IntegrationAuditLogPage.test.tsxapps/web/src/lib/ai/core/vision-models.ts
- Add qwen3-max-thinking to vision model registry (Codex + CodeRabbit) - Add grok-4-fast to vision model registry (CodeRabbit) - Add GPT-5.4/5.4 Pro to direct OpenAI provider (CodeRabbit) - Fix TypeScript error: cast Select onValueChange to FiltersState union
The tasks route now imports checkMCPPageScope from @/lib/auth, so the existing mock needs to include it to avoid vitest errors.
2 participants
Summary
Add new model support across Anthropic, Google, OpenAI, and OpenRouter providers, plus integration system enhancements.
AI Models
Integrations
Infrastructure
How to Validate