Google ships Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking audio-to-audio models
2026-09-15
Google's Gemini API changelog announced general availability of Gemini 3.8 Live (default low-latency voice agent model with interleaved reasoning and asynchronous function calling) and Gemini 3.8 Live Extended Thinking (a high-reasoning variant supporting background reasoning during live audio interactions), both real-time voice models using the Live API.
Significance 3: A real GA release extending Google's real-time-agent surface, but an incremental voice-niche capability rather than a frontier reasoning or coding move.
Implications · machine-drafted, not owner judgment
GA audio-to-audio voice models with interleaved reasoning extend Google's real-time-agent surface, but this is an incremental capability release in a niche (voice) rather than a frontier reasoning or coding move, so it doesn't move any tracked verdict on its own.
- Independent latency/quality comparisons against OpenAI's or xAI's voice offerings
Sources
- primary Gemini API release notes retrieved 2026-09-17