Gemini 3.5 Live Translate turns real-time speech translation into a developer surface
Google’s Gemini 3.5 Live Translate is rolling into the Live API, AI Studio, Translate, and Meet, with 70+ language detection and SynthID-watermarked audio.
Google’s Gemini 3.5 Live Translate is the kind of AI feature that makes more sense as infrastructure than as a demo: it listens continuously, detects more than 70 languages, and streams translated speech a few seconds behind the speaker while trying to preserve intonation, pacing, and pitch.
The model is now available to developers in public preview through the Gemini Live API and Google AI Studio under gemini-3.5-live-translate-preview. Google is also moving the same capability into consumer and Workspace products: Google Translate is rolling it out globally on Android and iOS, while Google Meet starts with a private preview for select business Workspace customers this month and a broader rollout later this year.
Technically, this is not just phrasebook translation. Google says generated audio is watermarked with SynthID, Meet translation expands from five languages to more than 70 languages and more than 2,000 language combinations, and the Live API examples use a stateful WebSocket flow with 16 kHz PCM audio input and 24 kHz PCM audio output.
Developers can try the model directly in AI Studio or use Google’s Gemini Live API examples on GitHub, including a LiveKit sample for real-time media sessions. Google also says partners including Agora, Fishjam, LiveKit, Pipecat, and Vision Agents are integrating with the Gemini Live API, which matters because the hardest part of live translation is often the audio transport and room infrastructure around the model.
The competitive read is straightforward: Google is turning its consumer-scale translation advantage into a programmable model surface before live interpretation becomes just another meeting, classroom, support, and travel primitive. OpenAI, Microsoft, and Meta all have speech translation assets, but Google can distribute this across Translate, Meet, Android, AI Studio, and the Gemini API at once.
Google did not publish pricing, quota, session length, SLA, or regional details for gemini-3.5-live-translate-preview in the launch post. For now, treat the API as public preview, Meet as private preview, and Translate as a rolling app feature.