Gemini 2.0 Flash: natively multimodal with audio and image output

Gemini 2.0 Flash is not just a model that understands images and audio: it's the first Google model capable of producing output in all modalities natively. You can talk to it, show it real-time video, and it responds with voice, text, and images generated autonomously. The 50-millisecond per token latency makes it usable in natural real-time conversations. It also integrates tools like web search and code execution, making it a true multimodal agent.