| name | update-google-model |
| description | Add a new Google model version (Gemini or Gemma) to pidgin/data/models.json. Use when the user mentions a newly released gemini-* (gemini-3.0-pro, gemini-2.5-flash, etc.) or gemma-* model. |
Update Google model
Follow the universal recipe in update-model, with these Google-specific details.
Conventions
- Key: the canonical Google model id — e.g.
gemini-3.0-pro, gemini-2.5-flash, gemma-3-27b-it.
- display_name: existing entries use the bare id (
"gemini-2.5-pro"). Match that style.
- api.model_id: equal to the key.
Family aliases
| Family / tier | Alias |
|---|
| Latest Gemini Pro | gemini |
| Latest Gemini Flash | flash |
| Previous-major Flash | flash<MAJOR> (e.g. flash2 for gemini-2.0-flash) |
When bumping a major Pro or Flash release, move the alias to the new entry and clear the previous owner's aliases to []. Gemma models typically have no alias.
Pricing & limits
Gemini pricing varies (Pro vs Flash). Copy from the closest sibling and flag in the summary so the user can confirm. max_context_tokens for current Gemini 2.x sits in the 1M range — copy from the sibling rather than guessing.
Capabilities
Copy from the closest sibling. Gemini 2.5 Pro and Flash have vision: true in the current data.
Recipe
- Locate the closest sibling (same tier — Pro/Flash/Gemma variant).
- If the new model takes over
gemini or flash, clear that alias from the previous owner.
- Insert the new entry cloned from the sibling, with new key,
display_name, api.model_id, appropriate alias(es), and today's cost.last_updated.
- Validate and commit per
update-model.
Commit message format: chore: add <new-key> (add and move <alias> alias if an alias was moved).
