Gemini Live receives its biggest upgrade yet. With the new model, conversations will be smoother

Those who regularly use the Gemini Live voice assistant may have already noticed that conversations with it are suddenly flowing a little more smoothly. Google has deployed a brand new model under the hood, Gemini 3.1 Flash Live, which the company calls the highest quality voice model to date in its portfolio. The change is primarily reflected in the speed and naturalness of the dialogue.

More natural conversation without awkward pauses

The main new feature is significantly lower latency, which allows responses to come more smoothly and without those awkward delays that often accompany voice assistants. Furthermore, the 3.1 Flash Live model can follow the conversation thread twice as long as its predecessor, so it won’t lose track as easily during longer brainstorming sessions.

Gemini Live just got its biggest upgrade yet, powered by Gemini 3.1 Flash Live.

•Faster responses with fewer awkward pauses
•Smarter & able to follow along 2x longer conversations, so you can stay in the flow
•Dynamically adjusts its answer lengths & tone to match the moment pic.twitter.com/b4YaJi3W7a

— Google Gemini (@GeminiApp) March 26, 2026

Google also emphasizes a better understanding of voice tone. The new model reportedly recognizes when a user is frustrated or confused, and accordingly adjusts not only the content but also the length and style of its response. In the ComplexFuncBench Audio benchmark, which tests the ability to perform complex tasks, the model achieved a score of 90.8%.

Already available in the mobile app

The improved Gemini Live is now available in the Gemini mobile app for both Android and iOS. Developers can test the new audio model via the Gemini Live API in Google AI Studio.

All audio generated by the 3.1 Flash Live model is automatically watermarked with SynthID, which allows for the detection of AI-generated content.

Have you noticed a difference in Gemini Live?

Sources: Google, Gemini/X