Update model definitions for latest OpenAI, Gemini, and Claude models#341
Update model definitions for latest OpenAI, Gemini, and Claude models#341williamjameshandley wants to merge 3 commits into
Conversation
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
![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: 9810b0e6ce
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".
| gemini-3.1-flash-live-preview: | ||
| # Model metadata | ||
| description: "Real-time audio-to-audio model for Live API with low-latency streaming" | ||
| capabilities: "🔴 Best for: Real-time audio conversations, low-latency voice interaction, Live API" |
There was a problem hiding this comment.
gemini-3.1-flash-live-preview is now exposed as a regular Gemini model, but our Gemini adapters only use client.models.generate_content(...) and do not implement a Live API/realtime session path. Since this model is explicitly a Live API audio-to-audio model, selecting it through normal mcp-chat generation will hit a runtime API error instead of returning output. Please either hide this model from standard generation/image-analysis routing or add a dedicated realtime adapter before registering it here.
Useful? React with 👍 / 👎.
- Add missing supports_vision: true to gpt-5.4-nano - Split gemini-3.1-flash-live-preview output pricing into text/audio - Fix stale usage note: Sonnet 4.5 → Sonnet 4.6 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
1 participant
Summary
gpt-5.4-mini(cheaper/faster GPT-5.4, ~1.05M context, reasoning/vision/caching),gpt-5.4-nano(cheapest/fastest GPT-5.4, API-only), andgpt-5.3-codex-spark(faster Codex variant, research preview, >1000 tok/s)gemini-3.1-flash-live-preview(real-time audio-to-audio model for Live API, low-latency streaming)opusalias fromclaude-opus-4-5-20251101toclaude-opus-4-6, andsonnetalias fromclaude-sonnet-4-5-20250929toclaude-sonnet-4-6Files changed
src/mcp_handley_lab/llm/providers/openai/models.yaml— 3 new model entriessrc/mcp_handley_lab/llm/providers/gemini/models.yaml— 1 new model entrysrc/mcp_handley_lab/llm/registry.py— updated 2 alias mappingsTest plan
opusandsonnetaliases resolve to the new model IDs🤖 Generated with Claude Code