Gemini 3.8 Live Extended Thinking

The high-complexity half of Google's live dialogue pair, released the same day as Gemini 3.8 Live. It adds multi-step reasoning to real-time conversation, and can explain its thinking while it works rather than going silent.

Best for

• Voice agents that must reason, not just answer
• Complex multi-step tasks driven entirely by speech
• Cases where a user needs to hear why the assistant is doing something

How it compares — and the India angle

• Reasoning out loud during a live call is the genuinely new behaviour here — the usual failure of voice assistants on hard questions is a long silence, which a caller reads as a dropped line.
• Reach for plain Gemini 3.8 Live when the job is conversation and this one when the job is work.
• Reaches further than its sibling on consumer surfaces: Gemini Live, and for Google AI Pro and Ultra subscribers inside Docs, Gmail and Keep.
• As with Gemini 3.8 Live, Google had not published per-token pricing at the time of writing, so none is quoted.