Google ships Gemini Live Translate

Google introduced real-time simultaneous interpretation via Gemini 3.5 'Live Translate.' The signal centers on official DeepMind sources with community (simon_willison) and arXiv involvement—an official announcement drawing developer-community reaction. The focus is a cross-conversation interpreting experience that translates speech back with low latency; the through-line is a shift from a translation-accuracy race toward the experience value of real-time responsiveness. It reads as multimodal models extending from 'read and write' to 'listen and translate on the spot.' But this is mainly a feature announcement—real-use latency and accuracy, language coverage, and the gap versus existing interpreting await validation.