Breaking

Google Unveils Gemini 3.5 Live Translate, Covering 70-Plus Languages by Voice

June 9, 2026 at 11:46 EDT

Foundation Models
Sales & Business

On June 9, 2026, Google announced "Gemini 3.5 Live Translate," a new audio model that translates speech to speech in near real time across more than 70 languages. It will roll out through three channels: the Google Translate app, a developer API, and Google Meet.

June 9 · Google DeepMind

Gemini 3.5 Live Translate: Real-Time Speech, As You Speak

An end-to-end audio model that starts interpreting the moment a speaker begins talking — streaming low-latency, natural-sounding translation across 70+ languages while preserving the speaker's own voice.

70+

Languages supported

2,000+

Language pairs

~sec

Few-second delay, live streaming

Direct audio vs. the old cascade

Fewer steps from voice to voice means lower latency and a more natural result.

Speech → Text

Text translation

Text → Speech

Cascade · 3 stages

Audio → Audio
(one model)

End-to-end · 1 stage

Keeps the speaker's own voice

Tone Pacing Intonation Pitch

Works in noisy environments and auto-detects mixed-language input.

Positive reception

"Natural enough to forget it's there." Praised for low latency and seen as a turning point for breaking language barriers — with hopes for travel, global meetings and software interpretation.

Still unknown

Exact latency figures, benchmarks, pricing and context length undisclosed. Supported devices, regions and privacy remain to be verified in real-world use.

Phased rollout

Translate app (iOS / Android) → AI Studio · API preview → Google Meet (soon)

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.