Google launches Gemini 3.8 Live and Extended Thinking voice models
Google announced on September 15, 2026 the release of two speech‑to‑speech models—Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking—designed for real‑time voice agents. The models support near‑real‑time visual grounding, automatic detection of 97 languages, and background tool or API calls that keep the conversation flowing. Gemini 3.8 Live Extended Thinking topped Artificial Analysis’…
Key points
- Google released Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking on Sep 15 2026 for real‑time speech‑to‑speech voice agents.
- Extended Thinking topped Artificial Analysis’ Speech‑to‑Speech Quality Index with 82.6 and scored 68.6% τ‑Voice, 35.1% τ‑Voice‑banking, 97.7% Big Bench Audio.
- Marktechpost reports pricing of $0.005/min input and $0.018/min output; Socialmediaexaminer alone notes a Windows desktop app and Nano Banana/Gemini Omni features.
Marktechpost alone provided pricing details of $0.005 per minute of audio input and $0.018 per minute of audio output, and added LangChain to the partner list. Socialmediaexaminer was the only source to mention a dedicated Windows 10/11 desktop app accessed via Alt + Space and to highlight additional Gemini features called Nano Banana and Gemini Omni. The ServiceNow EVA‑Bench benchmark was reported only by unite.ai.
Model pages: Gemini 3.8 Live → · Gemini 3.8 Live Extended Thinking →
The story so far
4 episodes →- Google launches Gemini 3.8 Live and Extended Thinking voice modelsthis story
Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking
Google DeepMind · 15 September 2026
Loading the full article…
This text was published by Google DeepMind and written by Tom Ouyang. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
7sources- Hacker News discussion · 54 pointsnews.ycombinator.com
- New home for CoworkNewsletter · Ben's Bites ·
- Creating AI Image Libraries, Showing Up in AI Results, and Industry NewsPress · Social Media Examiner ·
- Gemini Live audioNewsletter · Simon Willison ·
- Google Releases Gemini 3.8 Live and 3.8 Live Extended Thinking for Production Grade Voice AgentsPress · MarkTechPost ·
- Google launches Gemini 3.8 Live to take on OpenAI's GPT-Live-1 at a fraction of the costPress · The Decoder ·
- Google Launches Gemini 3.8 Live and Extended Thinking Voice ModelsPress · Unite.AI ·
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Anthropic releases Claude Haiku 5.5 at lower prices than GPT-6 Luna · 1 src
- Microsoft releases Microsoft-Decision-1, a Qwen3.5-9B-based scoring model · 2 src
- Mistral releases Mistral Large 4, nicknamed Le Chonk, a 1-trillion-parameter multimodal model · 1 src
- Google rolls out Gemini 4 Argon with new safety safeguards · 1 src
- Google launches Gemini 4 Argon model · 11 src
Comments
via GitHub Discussions