Google released Gemini 3.8 Live and 3.8 Live Extended Thinking voice models

On September 15, 2026, Google made two real-time voice models generally available, Gemini 3.8 Live and 3.8 Live Extended Thinking, with configurable thinking levels, 97 languages and per-minute pricing.

Archive entry: this covers an event from September 15, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On September 15, 2026, Google made two real-time voice models generally available, as announced in Google's post on Gemini 3.8 Live and 3.8 Live Extended Thinking: gemini-3.8-live and gemini-3.8-live-extended-thinking.

What changed

Thinking levels were MINIMAL, LOW, MEDIUM and HIGH, and the models supported asynchronous function calling.
Gemini 3.8 Live Extended Thinking ranked #1 on the Artificial Analysis Speech-to-Speech Quality Index with 82.6.
Extended Thinking scored 68.6% on τ-Voice, 35.1% on τ-Voice-banking and 97.7% on Big Bench Audio, and Gemini 3.8 Live ranked 2nd in Speech Agent Arena.
The models supported 97 languages and could switch language mid-conversation.
Pricing was $0.005 per minute of audio input and $0.018 per minute of audio output.
All generated audio was watermarked with SynthID.

What it means for builders

Developers got production voice agents with tunable reasoning at per-minute pricing, with audio input and output priced separately. Two models were available: Gemini 3.8 Live, and Gemini 3.8 Live Extended Thinking, which led the Artificial Analysis speech-to-speech index.

Availability and sources

Both models powered Search Live, Gemini Live and Live features in Docs, Gmail and Keep. Enterprise availability was in private preview. Developer details are in the Gemini API changelog entry and in a Google post on building real-time voice applications.

Sources

Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking (Google (The Keyword), September 15, 2026, official)
Gemini API changelog: September 15, 2026 (Google AI Studio (Gemini API changelog), September 15, 2026, official)
Build real-time voice applications with Gemini 3.8 Live and 3.5 Transcribe (Google (The Keyword), September 15, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.