Stocks
Google's Advanced Audio Models Top Analysts' Speech-to-Speech Index
Google has released its most advanced audio models, Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, which can talk, reason, and handle tasks.
The two models were released on September 15, 2026, by Google DeepMind for natural and production-ready voice applications.
According to Artificial Analysis' Speech-to-Speech Index, the Extended Thinking version topped the list with a score of 82.6, beating GPT-Live-1 and Grok Voice.
The standard Gemini 3.8 Live model placed fifth with a score of 76.0, while both variants beat the previous generation's Gemini 3.1 Flash Live High model, which scored 71.5.