DEV Community

Cover image for BREAKING: Google Launches Gemini 3.8 Live, Its Smartest Voice Dialogue Model Yet
Anas Hamad
Anas Hamad

Posted on Originally published at techmeme.com

BREAKING: Google Launches Gemini 3.8 Live, Its Smartest Voice Dialogue Model Yet

BREAKING: Google Launches Gemini 3.8 Live, Its Smartest Voice Dialogue Model Yet

Breaking straight from Google.

They just dropped two new models: Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, both built to make voice agents feel far less robotic.

Here's the simple version. The standard Live model is built for speed, it answers you almost instantly, like a friend who never hesitates before replying.

The Extended Thinking version takes a tiny beat longer on purpose, it actually 'thinks' before responding, especially for trickier questions that need real reasoning.

Why does this matter? Because the biggest flaw in voice AI has always been this tradeoff: fast responses that sound dumb, or smart responses that come with awkward latency.

Google is essentially trying to teach its models when to react instantly and when to pause and reason, just like a human would on a real phone call.

If this actually works the way they're promising, we're getting closer to a world where you call customer support and genuinely can't tell if you're talking to a person or a model that just quietly out-thought you.


🔗 Original Source & Reference: https://www.techmeme.com/260915/p41#a260915p41

Published automatically via FeedMind AI Content Pipeline.

Top comments (0)