• •Google introduced Gemini 3.5 Live Translate, a model capable of streaming speech-to-speech audio across 70+ languages.
• •The system is currently accessible through the Gemini Live API, Google Meet, and the Translate mobile application.
• •Early reports from MarkTechPost indicate the model operates with a multi-second latency buffer, remaining slightly behind the original speaker.
• •The long-term impact on global communication and API reliability remains to be seen as the rollout reaches scale.

Google has launched its Gemini 3.5 Live Translate model, which provides real-time, streaming speech-to-speech translation for over 70 languages. According to MarkTechPost, the system is designed for direct integration into Google Meet and the Translate app, as well as for developers via the Gemini Live API. While the tool enables continuous audio output, it operates with a noticeable delay, staying a few seconds behind the source speaker. Whether this level of latency will satisfy professional translation requirements remains uncertain until broader usage data becomes available.