Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking

Google has unveiled Gemini 3.8 Live and 3.8 Live Extended Thinking, two new AI models designed to transform voice interactions with devices. These advanced models enable seamless, natural, and intelligent conversations, handling interruptions, multiple languages, and background task execution without disrupting the dialogue.

Available through the Gemini API, Google Workspace, and the Gemini app, these models empower developers and enterprises to build reliable, production-ready voice agents. Gemini 3.8 Live Extended Thinking, in particular, excels in complex reasoning and real-time visual context, outperforming competitors on key benchmarks. It also offers strong cost-effectiveness compared to other frontier models.

Gemini 3.8 Live, another highlight, shows high user preference and remains cost-effective, processing visual inputs in near real-time and automatically handling language transitions mid-conversation. It can execute tools and API calls in the background while maintaining fluid conversations, providing progress updates as needed.

Google's Live models also power intuitive, collaborative experiences across Google Workspace and Search, particularly for complex tasks. They are integrated with various developer platforms, allowing creators to focus on crafting user experiences while handling real-time media streaming infrastructure. All audio generated by these AI products is watermarked with SynthID to prevent misinformation.