Gemini 3.5 Live Translate Preview

Gemini the listed price Live Translate Preview is purpose-built as a low-latency, audio-to-audio model that focuses on real-time translation of spoken conversations. Rather than serving as a general-purpose large language model, it is engineered to keep spoken exchanges flowing naturally, handling continuous input speech and returning translated audio alongside a matching text transcript. The result is a tightly scoped conversational translator where the model's primary job is to listen, translate, and speak back fluidly rather than to generate open-ended writing or reasoning.

In practical terms, this preview is best suited to live multilingual scenarios such as interpreting meetings, cross-language customer calls, or travel conversations, where minimal delay and a natural-sounding translated voice matter more than deep reasoning. It pairs incoming speech with outgoing translated speech for seamless bidirectional dialogue, while the returned transcript gives developers a reliable handle for downstream processing, logging, or display. Because it is offered as a preview under the broader Gemini family, it gives teams a way to prototype real-time voice translation experiences against a Google-maintained endpoint before committing to a production deployment.