Google has announced Gemini 3.1 Flash Live, calling it its “highest-quality audio and speech model” to power major upgrades to Gemini Live and Search Live. The model is now available in preview in Google AI Studio via the Gemini Live API. Compared with 2.5 Flash Native Audio, it offers lower latency and is better at recognizing subtle acoustic cues in speech such as pitch and speaking rate.
In complex environments, Gemini 3.1 Flash Live can better separate a user’s voice from background sounds like traffic or TV audio, with Google highlighting stronger background noise filtering. For Gemini Live on Android and iOS, the new model delivers faster responses and fewer awkward pauses, and it can extend how long it continuously tracks conversational context to twice the previous duration—supporting longer discussions and brainstorming. It also dynamically adjusts response length and tone to better match the conversation.
