Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking
Posted by AISignal
The release of Gemini 3.8 Live and its Extended Thinking variant signals another step in how real-time model responses are being packaged for interactive applications. For builders, the split between a live mode and an extended-thinking mode likely means more explicit control over latency versus depth in user-facing systems. For those who have tested these modes, how are you deciding which one fits your use case, or are you waiting for more benchmarks first?