Llm-gemini

- New [Gemini 2.5 Pro Experimental](https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/) model: `llm -m gemini-2.5-pro-exp-03-25`. #61
- `llm gemini models` command now returns a JSON array, not an object.
- New `llm gemini files` command returns a JSON array of files currently stored using the Gemini files feature. 19

- New [Gemma 3](https://blog.google/technology/developers/gemma-3/) model `gemma-3-27b-it`. #58
- `llm gemini models` command to see a JSON list of models currently available through the Gemini API.

- Fix for bug where `title` was not a valid property name. 55

- Support for new [gemini-embedding-exp-03-07 embedding model](https://developers.googleblog.com/en/gemini-embedding-text-model-now-available-gemini-api/), plus `gemini-embedding-exp-03-07-2048`, `gemini-embedding-exp-03-07-1024`, `gemini-embedding-exp-03-07-512`, `gemini-embedding-exp-03-07-256`, `gemini-embedding-exp-03-07-128` length variants. #53

- Fixed bug where `model.prompt(schema=Model)` failed to work for Pydantic models. 49
- Reduced the verbosity of the JSON recorded in `response.response_json`. 51

- Support for [schemas](https://llm.datasette.io/en/stable/schemas.html) using the new `llm prompt --schema X` and `model.prompt(..., schema=X)` mechanisms. #49