From Mexico — Google just introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking: its latest live dialogue models for voice agents that can keep talking while tools and deeper reasoning run in the background. Same-day developer detail is on the Gemini Audio / Live API post.
I’m reading this as a speech-to-speech product ship, not a funding headline. Google positions 3.8 Live for scale and cost-efficient fluid dialogue with visual grounding, and 3.8 Live Extended Thinking for high-complexity, multi-step work that still has to sound like a live conversation.
What Google says ships today
Per the company blog, 3.8 Live processes visual inputs in near real time, auto-detects and switches among 97 supported languages mid-conversation, and executes tools / API calls in the background while it keeps chatting. Extended Thinking is framed as reasoning and speaking at the same time — early verbal cues like “Let me check that…” plus live progress narration on multi-step background tasks.
Google’s own scoreboard (I’m not inventing third-party bake-offs): Extended Thinking takes the #1 overall spot on Artificial Analysis’ Speech to Speech Quality Index at 82.6, and Google cites 68.6% on τ-Voice and 35.1% on Sierra’s τ-Voice-banking for agentic task completion, plus 97.7% on Big Bench Audio. It also says 3.8 Live placed second in the Speech Agent Arena among users.
Developer path and pricing (Google’s numbers)
The developer post says both models are available via the Live API in the Gemini API and Google AI Studio, with async function calling, visual context, alphanumeric precision, multilingual coverage, and incremental content updates. Google lists Live API integration partners including Agora, Fishjam, LiveKit, LangChain, Pipecat, Vercel, and Vision Agents.
Listed price on that post: $0.005/min audio input and $0.018/min audio output (Google’s estimate footnote ties that to $3/$12 per 1M tokens). Treat that as Google’s published rate card, not an independent audit.
Where you can try it — tiered rollout
Google: 3.8 Live starts today for developers in the Gemini API and AI Studio; enterprises in private preview in Gemini Enterprise (Customer Experience “coming soon”); and for everyone in Search Live. Extended Thinking likewise hits the API / AI Studio today; private preview in Gemini Enterprise (plus Workspace business customers “coming soon”); and for everyone in Gemini Live, with Google AI Pro/Ultra subscribers getting Workspace Docs Live, and all Google AI subscribers getting Gmail Live and Keep Live.
One safety note from the same blog: audio from Google’s AI products is watermarked with SynthID.
My takeaway
From Mexico, the narrow read: Google shipped two Live speech-to-speech models today — 3.8 Live for fluid, tool-using dialogue, and Extended Thinking for heavier multi-step work that still narrates progress out loud — with a tiered rollout across API, Search Live, Gemini Live, and Workspace/Enterprise previews. Sources: Google blog, Gemini Audio developer post.
