Google Launches Gemini 3.1 Flash Live: Real-Time Voice AI Transforms Search Conversations

Google rolled out Gemini 3.1 Flash Live, a voice AI with major upgrades in speed, task completion, and realism across Search and API.

Google Launches Gemini 3.1 Flash Live: Search Enters Real-Time Voice AI Era

Google released Gemini 3.1 Flash Live, integrating real-time voice conversation into Google Search, Gemini app, and Android. Key breakthroughs: sub-200ms end-to-end latency for natural conversation pacing; ~35% improvement in complex query task completion; better handling of interruptions, context switches, and implicit intent.

This accelerates search's transition from keyword-and-links to conversational Q&A. For Google it's both defensive (countering ChatGPT/Perplexity) and offensive (leveraging search data advantages). Voice search creates new ad model challenges since traditional search ads depend on users browsing result pages.

Google also opened Flash Live APIs at ~1/10 Ultra pricing, enabling third-party voice AI integration for customer service, voice assistants, and in-vehicle systems. A dedicated voice conversation evaluation benchmark was also released.