Google Unveils Gemini Models to Boost Voice Agent Capabilities

Alphabet Inc., parent company of Google, has released two new Gemini models designed to enhance the conversing capabilities of voice agents. The 3.8 Live and 3.8 Live Extended Thinking models are aimed at developers and enterprises looking for reliable production-ready voice solutions.

The announcement was made on September 15, 2026. According to Google, these new models deliver improved building blocks for voice agents, allowing them to engage in more natural conversations with users.

For developers and enterprises, these models are designed to be production-ready, meaning they can be used in real-world applications without significant modifications. This is a key development in the field of artificial intelligence and voice recognition technology.