What changed
Google introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking in a post dated September 15 and updated September 17. It positions regular Live for conversational tasks at scale and Extended Thinking for more complex, multi-step work.
The announcement describes visual grounding and background tool execution while the conversation continues. Extended Thinking can reason and speak concurrently. The examples and performance comparisons come from Google and the evaluations it cites; this note does not claim that SaveMyToken has tested latency, accuracy, or production reliability.
What it means for your workflow
SaveMyToken editorial take
Evaluate a voice agent on both task completion and conversational timing. Fast speech alone does not tell you how long the underlying work takes.