Google released Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking for fluid voice conversations. These models allow AI agents to handle complex tasks and reasoning in near real-time.
The Gemini 3.8 Live model provides multilingual support and high-scale performance for Google Search Live. The Extended Thinking version manages multi-step tasks requiring deeper reasoning.
Google reports these models achieved top industry rankings for speech-to-speech quality. The models also lead benchmarks for agent-based task completion.