Google has launched Gemini 3.1 Flash Live, a new audio model that the company describes as its most 'natural' and 'reliable' yet. Google promises that voice assistants will now sound human, moving beyond robotic responses that might evoke science fiction scenarios. The model boasts enhanced speed and rhythm, aiming to make user interactions more intuitive and less frustrating for people communicating with AI.
Google's performance claims for Gemini 3.1 Flash Live are supported by specific benchmarks. The model achieved a score of 90.8% on the ComplexFuncBench Audio benchmark, which simulates complex dialogue scenarios. Additionally, it scored 36.1% on Scale AI’s Audio MultiChallenge, a test where the model was trained to understand commands in real-world conditions and incorporated a 'thinking' function.
For businesses, this development signals potential cost savings in customer service operations. Gemini 3.1 Flash Live is integrated into Gemini Enterprise for Customer Experience. This means companies currently investing heavily in customer support can offload a portion of their workload to AI. Early adopters like Verizon and The Home Depot have reportedly praised the new model for its 'natural' sound and are planning to expand its use across multiple languages.
This initiative underscores Google's serious commitment to advancing voice interfaces. The company's objective is to reduce customer support costs by deploying AI to handle tasks traditionally performed by humans. The early results suggest this strategy is proving effective. AI assistants are poised to become faster, more accurate, and, critically, more pleasant to interact with, particularly for users needing quick solutions in noisy environments or while on the move.