Google releases Gemini 3.8 Live and 3.8 Live Extended Thinking

The models enable real-time reasoning, background tool execution, and mid-conversation language switching across 97 languages for voice agents.
Key points
- Gemini 3.8 Live Extended Thinking reasons and speaks simultaneously, using filler cues and progress narration during multi-step tasks.
- Gemini 3.8 Live supports near real-time visual input processing and background API calls while maintaining ongoing dialogue.
- Both models are available through the Gemini API and Google AI Studio, and are rolling out to Search Live, Gemini Live, and Google Workspace.
- All generated audio outputs include imperceptible SynthID watermarks.
Why it matters
Developers building conversational AI can deploy real-time voice agents that execute complex background workflows without pausing dialogue.
Sources
- Google Releases Gemini 3.8 Live Dialogue Models (Google AI Studio, Sep 15, 2026, 5:38 PM)
- Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking (Google DeepMind, Sep 15, 2026, 5:09 PM)