Google launches Gemini 3.8 Live voice AI models

Google introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking on Sept 15, 2026, its most advanced real-time voice AI models. They support 97 languages, visual input, parallel reasoning, and tool-calling. Available via Gemini API, Google AI Studio, Search Live, and Workspace, its partners include Salesforce, LangChain, and LiveKit.