Google Unveils Gemini 3.8 Live Extended Thinking for Real-Time Workshop Guidance and Complex Problem-Solving

Google has launched Gemini 3.8 Live Extended Thinking, a new voice-first model that reasons aloud while guiding users through complex, multi-step tasks in real time. The model, announced on September 15, 2026, combines extended thinking reasoning with live audio interaction, enabling applications like workshop assistants that analyze physical spaces, narrate their reasoning verbally, and provide step-by-step guidance throughout building or repair processes.