Google launched Gemini3.8Live and Gemini3.8Live Extended Thinking, native end-to-end voice models enabling ultra-low latency conversation with deep multi-step reasoning. The standard version optimizes natural real-time dialogue, while the Extended Thinking variant performs complex background processing during speech. These models excel in troubleshooting, coding, and tutoring, now available via APIs for enterprise and developer integration into smart hardware and customer service applications.