Google Launches Gemini 3.1 Flash Live: Real-Time Voice-First AI with 90+ Language Support and Workspace Integration

Google introduced Gemini 3.1 Flash Live as its most advanced voice-first AI model, featuring low-latency real-time voice interactions with improved acoustic nuance recognition, background noise filtering, and natural rhythm processing across 90+ languages. The model supports chat history import for personalized experiences and targets voice commerce, navigation, and hands-free agent scenarios, with cross-platform adaptability particularly suited for startups. Google simultaneously expanded Search Live and Gemini integration into Docs, Sheets, Slides, Drive, and Google Maps, broadening the Workspace AI footprint. Ethan Mollick's viral 'Three Years from GPT-3 to Gemini 3' post — in which Gemini 3 built a fully playable 'Candy-Powered FTL Starship Simulator' on prompt — has become a widely-cited proxy for measuring AI's three-year capability leap.