Add support for Google’s Gemini model family as an option in the char tool so users can select and run Gemini-based agents/characters in addition to existing models. This will provide higher-quality responses, multimodal capabilities (if needed), and parity with providers that already expose Gemini.

Why this matters / Benefits

Improved model capabilities: Gemini often yields better factuality, instruction-following, and creativity compared with some older LLMs.

Multimodal support: Enables future features that use images, audio, or structured inputs (if char tool later supports multimodal inputs).

Competitive parity: Allows users to compare/choose across providers and keeps the tool up-to-date with industry-leading models.