Google Releases Gemini 3.8 Live and Flash Featuring Extended Thinking, Avatars, and Custom Voice Narration
Context that changes how you build, even if there's nothing to install.
On September 24, 2026, Google officially released Gemini 3.8 Live, which adds live visual avatar functionality and advanced reasoning capabilities for real-time interactions.
It focuses on the 'human-like' interface of AI, which is critical for consumer apps but less relevant for core developer workflows.
Visual avatars are high-bandwidth gimmicks for the time being. While the 'Extended Thinking' part is interesting, the primary focus here is consumer engagement rather than tool intelligence.
One sentence: if these avatars can be integrated into third-party apps via a low-latency video API.
Also covers
- Google Launches Gemini 3.8 Flash with Custom Voice NarrationGoogle has updated the Gemini 3.8 Flash model to include advanced Text-to-Speech (TTS) features. Developers can now generate custom voice narration and real-time audio responses with higher fidelity and lower latency than previous versions. The update is being showcased in new 'Live' features in the Gemini app.
- Extended Thinking allows the model to process more complex queries before responding, improving accuracy for logic-heavy tasks.
- Live Avatars provide a more human-like interface for customer-facing agents and educational tools.
- The integration into vehicles represents a significant expansion of LLM utility into physical, daily-use environments.
Google is prioritizing low-latency, multimodal interactions across their 3.8 series models.