feat: add Gemini 3 Pro to Google provider model list (#405)#521
feat: add Gemini 3 Pro to Google provider model list (#405)#521mvanhorn wants to merge 1 commit into
Conversation
Adds gemini-3-pro to providers.google.models in api/config/generator.json with the same temperature/top_p/top_k shape as the existing 2.5 entries, and mirrors the addition in the fallback ModelConfig in api/api.py:220 so the new model surfaces in the dropdown even if config load fails. default_model remains gemini-2.5-flash so existing users see no behavior change. Users opt into 3 Pro explicitly via the model picker. Fixes AsyncFuncAI#405
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request adds support for the Gemini 3 Pro model by updating the model list in the API and providing its configuration in generator.json. Feedback was provided to reorder the models in the API fallback list to ensure consistency with the configuration file and prevent potential issues with default model selection in the frontend.
| Model(id="gemini-3-pro", name="Gemini 3 Pro"), | ||
| Model(id="gemini-2.5-flash", name="Gemini 2.5 Flash") |
There was a problem hiding this comment.
The order of models in the fallback list is inconsistent with the main configuration in generator.json. In the configuration file, gemini-2.5-flash is the first entry and the designated default model for the Google provider. Placing gemini-3-pro at the top of the list in this fallback path might lead to it being incorrectly selected as the default model by the frontend if the configuration fails to load. It is better to maintain a consistent order across both the success and fallback paths.
| Model(id="gemini-3-pro", name="Gemini 3 Pro"), | |
| Model(id="gemini-2.5-flash", name="Gemini 2.5 Flash") | |
| Model(id="gemini-2.5-flash", name="Gemini 2.5 Flash"), | |
| Model(id="gemini-3-pro", name="Gemini 3 Pro") |
|
Small follow-up: I think gemini-3-pro is already superseded by gemini-3.1-pro-preview in the latest Gemini API docs. We should probably switch to gemini-3.1-pro-preview (or include both for compatibility). Reference: https://ai.google.dev/gemini-api/docs/models/gemini-3.1-pro-preview |
2 participants
Adds
gemini-3-proto the Google provider model list per #405.Changes:
api/config/generator.json- newgemini-3-proentry inproviders.google.modelswith the sametemperature/top_p/top_kshape as the existing 2.5 entriesapi/api.py:217-221- mirror the addition in the exception-path fallbackModelConfigso 3 Pro still surfaces in the dropdown ifload_generator_config()raisesdefault_modelfor the Google provider stays atgemini-2.5-flash, so existing users see no behavior change. Users opt into 3 Pro explicitly via the model picker.Pattern matches PR #420 (Bedrock model additions) and the original 2.5 family additions in PR #318.
Fixes #405