♦ The most visible consequence of the launch for ordinary users is the expansion of Search Live – the function that allows searching by voice on Google, with the possibility of showing objects through the phone's camera (Google Lens) – in over 200 countries and territories. Google has launched Gemini 3.1 Flash Live, an artificial intelligence model specialized in real-time voice conversations, which the company describes as «the most powerful voice and audio model to date». The model powers two Google products: Gemini Live, the voice assistant available on Android and iOS phones, and Search Live, the function that allows users to search Google through voice conversation, now extended to more than 200 countries and territories, in more than 90 languages, according to the official announcement published on the company's blog. «Gemini 3.1 Flash Live offers a more intuitive experience for developers, businesses and everyday users,» claims Google in the official announcement. The launch comes less than a month after Google unveiled Gemini 3.1 Flash-Lite (March 3), a developer-oriented model optimized for speed and low cost, which launched at $0.25 per million input tokens (tokens, the units of text processed by AI models) — eight times cheaper than the Gemini Pro model, according to VentureBeat. Gemini 3.1 Flash Live improves on the previous model (2.5 Flash Native Audio) on several dimensions, according to Google and 9to5Google: Latency is lower – responses come faster, with fewer awkward pauses in conversation. The model can follow the thread of the conversation for twice as long, which allows extended brainstorming sessions without the assistant losing context. It better recognizes acoustic nuances – tone, rhythm of speech, expressions of frustration or confusion -, dynamically adapting the length and tone of its responses. ZF The model filters background noise (traffic, TV) more effectively and obtained a score of 90.8% on the ComplexFuncBench Audio test, an indicator of the ability to use external tools (applications, functions) during the conversation, according to Google. All audio content generated by the model is marked with SynthID, a Google technology that embeds an imperceptible digital watermark directly into the audio signal to enable detection of AI-generated content and prevent disinformation, according to the company's announcement. The most visible consequence of the launch for ordinary users is the expansion of Search Live – the function that allows you to search Google by speaking, with the possibility of showing objects through the phone's camera (Google Lens) – in more than 200 countries and territories, in all languages and locations where the AI mode of Google Search is available, according to Droid-Life. Gemini Live, the voice assistant, also gets an upgrade: conversations are faster, more natural and can take place for longer periods without losing context. The model is available in preview through the Gemini Live API in Google AI Studio, the platform for developers, and through Vertex AI, the cloud service for companies. Google has separately published a technical guide for building real-time conversational agents based on the new model, according to Google's developer blog. At the corporate level, Gemini 3.1 Flash Live is already being used by Verizon, the American telecommunications operator, and Home Depot, the largest DIY store chain in the US, for customer service centers, according to Droid-Life. Also on March 26, Google extended the Live Translate function – which translates conversations in real time through headphones – to iOS (so far only available on Android). The feature supports over 70 languages and has been expanded to France, Germany, Italy, Japan, Spain, Thailand and the United Kingdom. It works with any standard headset, according to Droid-Life. The release of Gemini 3.1 Flash Live comes amid intense competition between Google, OpenAI and Anthropic in the artificial intelligence model segment, with each company releasing updates at weekly intervals. Google released three models of the Gemini 3.1 family in March 2026 alone: Flash-Lite (March 3, focused on speed and cost), Flash Live (March 26, focused on audio conversations) and, separately, updates to the flagship model Gemini 3 Flash.
For other news, analyses, articles and business information in real time, follow NV on WhatsApp Channels