Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking
If you use Gemini voice mode for hands-free tasks - like talking through a problem while driving or in a meeting - these models let it keep reasoning or working on your request in the background while you keep talking, instead of making you wait or repeat yourself.
What changed
Source summary
Google launched Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, new real-time voice models. 3.8 Live focuses on cost-efficient, fluid conversation with real-time visual understanding, while 3.8 Live Extended Thinking handles more complex reasoning tasks. Both can keep working on a task in the background without interrupting the conversation, and are available via the Gemini API, Google Workspace, and the Gemini app.
Why it matters to your work
Practical interpretation
If you use Gemini voice mode for hands-free tasks - like talking through a problem while driving or in a meeting - these models let it keep reasoning or working on your request in the background while you keep talking, instead of making you wait or repeat yourself.
When this could help
Worth evaluating if you are building a voice agent and need multilingual support, background tool execution during conversation, or real-time reasoning that does not pause the dialogue.
Gemini 3.8 Live is Google's cost-efficient live dialogue model supporting 97 languages with mid-conversation language switching and background tool calling. Gemini 3.8 Live Extended Thinking adds multi-step reasoning while speaking — it reasons and talks simultaneously, using verbal cues like 'Let me check that...' and live progress narration. Extended Thinking ranked #1 on the Artificial Analysis Speech-to-Speech Quality Index (82.6).
How to get started
Access both models through the Gemini Live API or Google AI Studio. Select the appropriate model based on whether you need cost efficiency (3.8 Live) or complex reasoning (Extended Thinking).
Provider guidance ↗
Build it into your routine
SCALA suggested workflow
Voice agent model comparison
Run a representative voice agent scenario through both 3.8 Live and Extended Thinking. Compare latency, reasoning quality, and cost to determine which model fits each use case in your pipeline.
What to check
Sources & provider guides
Official documentation checked 12 Oct 2026. SCALA prompts and workflows are suggestions, not claims of hands-on testing.
Source & availability
This is a summary of a published provider announcement, with practical context from SCALA’s feed.
Plan, region, workspace and rollout conditions may affect access in Australia. Check the original source for current availability.
blog.google