Gemini 3.1 Flash Live targets real-time voice and vision agents
- Source
- x.com
- Date
Listen up 🔊Gemini 3.1 Flash Live is launching today, making a big difference for developers who are building real-time voice and vision agents. How, you ask? Well, this model delivers: — Responses that feel as fast as natural dialogue — Better task completion in noisy environments — Improvements in complex-instruction following
Start building with 3.1 Flash Live anywhere it’s available: — Gemini Live in @GeminiApp + Search Live — Gemini Live API + @GoogleAIStudio in preview — Gemini Enterprise for Customer Experience And learn more about the model in our blog➡️ https://t.co/RmIDqJ4rP6
Context
Gemini 3.1 Flash Live is aimed at voice agents that must follow instructions and use software tools during a conversation. Google AI's launch article reports 90.8% on ComplexFuncBench Audio, a test of multi-step tool calls under constraints.
The accompanying chart compares that result with 71.5% for the December 2025 Gemini 2.5 Flash Native Audio version, both using high thinking effort. The test converted text prompts into audio, so the result specifically supports improved tool selection and execution from spoken instructions under those test conditions. Published comparison.
- postGoogle's week: EmbeddingGemma 2, Nano Banana 2.1, Gemini business agent
postGoogle opens SynthID watermark checks to everyone, with partner coverage
postGemma 4 orchestrates a plant DNA model to find crop mutations in minutes
articleGoogle's Project CC gives a family agent its own identity and sandbox
Checking sign-in…
Loading comments…