Sep 16, 2026 13:22:00

Google has released Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, voice-activated AI that enables real-time conversation while coding.

Google has released two AI models capable of real-time conversation: ' Gemini 3.8 Live ' and ' Gemini 3.8 Live Extended Thinking .' Gemini 3.8 Live Extended Thinking has achieved higher scores in benchmark tests than rival models from OpenAI and SpaceXAI.

Gemini 3.8 Live & Gemini 3.8 Live Extended Thinking

https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-8-live-gemini-3-8-live-extended-thinking/

New Gemini Audio models for developers
https://blog.google/innovation-and-ai/technology/developers-tools/build-real-time-voice-applications-gemini-audio/

The Gemini 3.8 Live is a cost-effective model, while the Gemini 3.8 Live Extended Thinking is a model with advanced intelligence and multi-stage reasoning capabilities. Both models support 97 languages, including Japanese , and can converse in real time.

In a demo video released by Google, you can see that it responds in real time according to the context, following the voice command 'Respond only when you say Gemini.'

What's new in the Gemini Live API - YouTube


A demo video has also been released showing how to code while interacting with Gemini 3.8 Live Extended Thinking.

Watch how we used 3.8 Live Extended Thinking to act as a programming tutor.

Both models feature:
🔵 Upgraded reasoning
🔵 Near real-time visual understanding
🔵 Automatic detection for 97 languages
🔵 Background tool calling without disrupting your chat

For your most difficult… pic.twitter.com/WHrNvqtHhl

— Google DeepMind (@GoogleDeepMind) September 15, 2026

The graph below shows task completion rate on the horizontal axis and conversation quality on the vertical axis. Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking achieve a balance between conversation quality and task execution accuracy.

In tests conducted by the third-party organization Artificial Analysis, Gemini 3.8 Live Extended Thinking outperformed OpenAI's GPT-Live-1-Astra and SpaceXAI's Grok Voice Think Fast 2.0.

Furthermore, in

Speech Agent Arena , where humans evaluate the quality of AI without knowing its name, Gemini 3.8 Live was rated as the second-best overall performer. The top-ranked AI was Gemini 3.1 Flash Live Minimal.

Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking are available via API. API documentation can be found at the following link.

Live API Feature Guide | Gemini API | Google AI for Developers
https://ai.google.dev/gemini-api/docs/live-api/capabilities?hl=ja#supported-languages