Google has officially announced the launch of the Gemini 3.1 Flash Live model, calling it the "highest quality audio and speech model" to date. This model delivers notable functional improvements to Gemini Live and Search Live services. Currently, developers can preview it through the Gemini Live API in Google AI Studio.
Compared to previous models, Gemini 3.1 Flash Live is more effective at recognizing sound details like pitch and speech rate, while also reducing latency. A key enhancement is its powerful background noise filtering capability, allowing it to more accurately distinguish relevant speech from ambient noises such as traffic or TV sounds. In the Gemini Live app on Android and iOS, the new model offers faster response times, reduces unnatural pauses, and doubles the memory duration for conversational coherence, aiding in maintaining focus during extended brainstorming sessions. Additionally, Gemini Live can now dynamically adjust response length and tone to match the dialogue context.
