Use Google Gemini Pro, Ultra, and Flash models for multimodal reasoning, long-context understanding, and fast inference. Supports text, image, and code generation tasks.