Google has expanded its Gemini 3.8 lineup with two voice-focused models designed to keep a conversation moving while AI tasks run in the background. Gemini 3.8 Live is positioned around speed and lower cost, while Gemini 3.8 Live Extended Thinking is intended for more complex, multi-step work.
Gemini 3.8 Live can call tools and APIs without pausing the dialogue, process visual context in near real time and automatically switch among 97 languages during the same conversation. The models follow the recent arrival of Gemini 3.8 Flash.
Extended Thinking adds background reasoning during a live exchange. Rather than going silent while it processes a request, the model can acknowledge that work is underway and provide progress updates while continuing to speak with the user. Google reports scores of 82.6 on the Speech-to-Speech Quality Index and 97.7% on Big Bench Audio for the model.
The approach moves Gemini Live beyond a conventional voice question-and-answer system toward an assistant that can continue a discussion while executing a task. Google has already brought voice interaction to productivity services including Docs, Gmail and Keep; Extended Thinking is intended to provide the more capable reasoning layer for those longer-running requests.
Rollout and Workspace access
- Gemini 3.8 Live is beginning to roll out in Search Live.
- Gemini 3.8 Live is beginning to roll out through the Gemini API.
- Gemini 3.8 Live is beginning to roll out in Google AI Studio.
- Gemini 3.8 Live Extended Thinking is arriving in Gemini Live.
- Google AI subscribers can access Extended Thinking in Workspace services including Gmail and Keep; Docs access is limited to Pro and Ultra plans.
Google says all generated audio is labeled with SynthID, its technology for identifying AI-generated material. How reliably Extended Thinking handles long conversations, interruptions and requests involving external services will be the more consequential test as the rollout expands.
