feat: add Gemini 3 Flash support #369
Conversation
Summary of ChangesHello @BjornMelin, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request introduces support for Google's recently launched Gemini 3 Flash model, Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
![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 new Gemini 3 Flash model. The changes are comprehensive, updating model configurations, documentation, and a large number of tests to reflect the new model and its aliases. The default fallback model is also updated to the new Flash model. I've found one area for improvement in the model selection logic within the Gemini provider to make it more robust. Overall, this is a well-executed and thoroughly tested feature addition.
There was a problem hiding this comment.
Pull request overview
This PR adds support for Google's newly launched Gemini 3 Flash model (gemini-3-flash-preview), which is positioned as offering Pro-level reasoning at Flash-level speed and cost. The changes update model aliases, configuration files, provider logic, tests, and documentation to accommodate this new model.
Key Changes:
- Updated
flashalias to point togemini-3-flash-previewinstead ofgemini-2.5-flash(higher intelligence score: 17 vs 10) - Added
flash2.5alias for legacy access togemini-2.5-flash - Updated default fallback model from
gemini-2.5-flashtogemini-3-flash-preview
Reviewed changes
Copilot reviewed 35 out of 46 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
conf/gemini_models.json |
Added Gemini 3 Flash model configuration with capabilities (1M context, 64K output, 32K thinking tokens) and updated aliases |
conf/openrouter_models.json |
Added Gemini 3 Flash support for OpenRouter with proper aliases and capabilities |
.env.example |
Updated documentation to list Gemini 3 Flash models and explain new alias structure |
providers/registry.py |
Changed default fallback model from gemini-2.5-flash to gemini-3-flash-preview |
providers/gemini.py |
Updated model selection logic to prefer canonical names and added documentation about thinking configuration |
docs/configuration.md |
Added Gemini 3 Flash to model table with updated aliases |
docs/advanced-usage.md |
Added Gemini 3 Flash documentation describing capabilities and use cases |
tests/test_thinking_modes.py |
Added new test class for Gemini-specific thinking configuration testing |
| Multiple test files | Updated test expectations to reflect new model defaults and alias changes |
| Various tool files | Applied PEP 8 formatting fixes for spacing around operators and string formatting |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
- Replace brittle string matching ("preview" or "-lite" suffix) with
registry-based approach using list_models(include_aliases=False)
- Update thinking config comment to be less speculative about future
SDK changes and Gemini 3 models
Addresses PR review feedback for more robust model name handling.
1 participant
Description
Adds support for Google's newly launched Gemini 3 Flash model (
gemini-3-flash-preview), which offers Pro-level reasoning at Flash-level speed and cost.Changes Made
gemini-3-flash-previewtoconf/gemini_models.jsonwith proper capabilities (1M context, 64K output, 32K thinking tokens)gemini-3-flash-previewtoconf/openrouter_models.jsonfor OpenRouter accessflashalias to point togemini-3-flash-preview(higher intelligence score: 17 vs 10)flash2.5alias for legacy access togemini-2.5-flashproviders/registry.pyfromgemini-2.5-flashtogemini-3-flash-previewproviders/gemini.pymodel selection to prefer canonical model namestests/test_thinking_modes.pyfor thinking mode validation.env.example,docs/advanced-usage.md, anddocs/configuration.mdModel Details
gemini-3-flash-previewflash,flash3,gemini-3-flash,gemini-flashTesting
Related Issues
N/A - Proactive update for newly released Gemini 3 Flash model
Checklist
feat: add Gemini 3 Flash support)Additional Notes
flashalias now points togemini-3-flash-previewfor best default experiencegemini-2.5-flashviaflash2.5alias