Gemini API I/O updates- Google Developers Blog

This post details several new models and functionalities for the Gemini API. New models include Gemini 2.5 Flash Preview (gemini-2.5-flash-preview-05-20) with improved reasoning, code, and long context, and enhanced cost-efficiency. Gemini 2.5 Pro and Flash text-to-speech (TTS) previews are available with native audio output for single/multiple speakers across 24 languages, offering control over expression and style. Gemini 2.5 Flash native audio dialog (preview) provides real-time conversational AI with proactive audio and emotional tone recognition. Lyria RealTime enables live music generation via WebSockets. Gemini 2.5 Pro Deep Think is an experimental reasoning mode for complex prompts. Gemma 3n is a new generative AI open model for everyday devices. New API functionalities include thought summaries for 2.5 Pro and Flash, thinking budgets for 2.5 Flash (extending to 2.5 Pro soon), an experimental URL Context tool, a computer use tool for browser control, broader JSON Schema support for structured outputs (including '$ref' and 'prefixItems'), video understanding improvements (YouTube URLs, uploads, clipping, dynamic FPS, multiple resolutions), asynchronous function calling for Live API, and a batch API for high-throughput, cost-effective processing.