Google DeepMind Launches Gemini 3.8 Live Real-Time Voice Models

Google DeepMind released two speech-to-speech models: Gemini 3.8 Live, positioned as a cost-effective conversational AI, and Gemini 3.8 Live Extended Thinking, designed for complex multi-step reasoning. Both models support real-time visual input, automatic language detection across 97 languages mid-conversation, and audio watermarking via SynthID. Gemini 3.8 Live Extended Thinking ranked first on Artificial Analysis' Speech to Speech Quality Index with a score of 82.6 and leads agentic task completion benchmarks, according to Google DeepMind. Gemini 3.8 Live secured second place in the Speech Agent Arena. Extended Thinking reasons and speaks simultaneously, narrating its process with verbal cues while executing tools and API calls in the background. The models are available to developers via Google AI Studio and target enterprises and consumer products, with named partners including ServiceNow, Agora, LangChain, LiveKit, Pipecat, Vercel, Salesforce, Genspark, and Lumeris.