Google releases Gemini 3.1 Flash Live with improved audio AI capabilities

Google on Thursday released Gemini 3.1 Flash Live, an updated model in its Gemini family that the company said delivers more natural and reliable real-time audio interactions.

The release targets improvements to live audio capabilities, an area of growing competition among major AI providers seeking to make voice-based AI assistants sound less robotic and more conversational.

Google said the new model version focuses on two core areas: naturalness in speech generation and reliability during extended audio sessions. The update builds on the existing Gemini Flash architecture, which is designed for speed and efficiency compared to the company’s larger Gemini Pro models.

The move comes as real-time audio AI has emerged as a key battleground in the industry. OpenAI launched its Advanced Voice Mode for ChatGPT last year, while Anthropic and other competitors have been developing their own voice interaction capabilities.

Flash Live sits within Google’s tiered Gemini model lineup, which ranges from the lightweight Flash series to the more capable Pro and Ultra variants. The Flash models are optimized for low-latency applications where rapid response times are critical, making them a natural fit for live audio use cases.

Google has been steadily expanding the Gemini family since its initial launch in late 2023, with successive updates aimed at closing performance gaps with competitors while maintaining the speed advantages of its smaller models.

The company did not immediately disclose detailed benchmarks comparing Gemini 3.1 Flash Live to its predecessors or competing models from other providers.

The update is available through Google’s AI platform, though the company has not yet specified pricing changes or whether the model will be integrated into consumer-facing products such as Google Assistant or the Gemini chatbot app.