Google fixed one of the problems that annoy users of its AI model

Google has announced the Gemini 3.8 Live and Extended Thinking models, which eliminate waiting times when working with voice assistants. The new model can think and speak simultaneously, using natural verbal cues. The system integrates with Google Workspace services, supports automatic switching between 97 languages, and achieved a record score in speech quality metrics.

Google announced last night (Tuesday) the launch of two new artificial intelligence models: Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking. The main innovation is the elimination of long waiting times when performing complex tasks with voice assistants. The Extended Thinking model can think and speak simultaneously, using natural verbal cues like 'let me check that for you,' and continues to converse with the user while running processes in the background. The system integrates with Google Workspace services: Gmail Live for voice search in the inbox, Docs Live for drafting documents, and Keep Live for creating notes. The model supports automatic switching between 97 languages mid-conversation. In performance tests, the Extended Thinking model took first place in the Artificial Analysis speech quality index with a score of 82.6. Google has implemented a digital watermark using SynthID technology to prevent fake news. The models are being deployed starting today in Google AI Studio, Gemini Enterprise, Search Live, and the Gemini app.