Gemini 3.8 Live

What changed

Google released two live dialogue models: Gemini 3.8 Live, built for scale and cost with near real-time visual input, and Gemini 3.8 Live Extended Thinking, which reasons and speaks at the same time for multi-step tasks. Both switch among 97 languages mid-conversation and run tools in the background. Google reports Extended Thinking ranks first on Artificial Analysis' Speech to Speech Quality Index at 82.6, scores 68.6% on tau-Voice and 97.7% on Big Bench Audio, and that 3.8 Live placed second in the Speech Agent Arena.

Access and availability

Both are stable in the Gemini API and Google AI Studio as gemini-3.8-live and gemini-3.8-live-extended-thinking. Consumers get them in Search Live and Gemini Live, and enterprises in a private preview of Gemini Enterprise.

Pricing

Paid tier, both models: text $0.75 per million input tokens and $4.50 per million output tokens; audio $3.00 per million input tokens ($0.005/min) and $12.00 per million output tokens ($0.018/min); image/video input $1.00 per million tokens ($0.002/min). A free tier is offered.

What you can build

Prototype a voice onboarding assistant that answers questions and walks new users through a multi-step setup.

Limitations

Enterprise access through Gemini Enterprise is still private preview, with Customer Experience and Workspace availability listed as coming soon. Google gave no numeric latency figures, and benchmark results are Google-reported.

Sources and release history

• Official Google announcement
• More releases from September 2026
• All Google releases