Gemini 3.8 Live and Live Extended Thinking
Gemini 3.8 Live and Live Extended Thinking are audio-to-audio models for voice agents, with 97.7% on Big Bench Audio and background reasoning mid-conversation.
Live is the low-latency default; Extended Thinking reasons in the background during a call. Second place in the Speech Agent Arena. 3.8 Flash TTS and Flash-Lite TTS with voice design and replication followed 2026-09-22.
- Date
- Tuesday, 15 September 2026
- Lab
- Kind
- model
- Access
- closed API
Figures
| Measure | Value | Measured by |
|---|---|---|
| Big Bench Audio | 97.7% Gemini 3.8 Live; company-reported | company |
Sources
- deepmind.google/blog/introducing-gemini-3-8-live-and-3-8-live-extended-thinking/
- ai.google.dev/gemini-api/docs/changelog
This record was checked against its sources on 6 October 2026. How we check