On 9 June 2026, Google announced Gemini 3.5 Live Translate. The model automatically detected more than 70 languages, preserved vocal characteristics and translated continuously rather than waiting for a completed turn.
This release brief was checked against first-party material on 10 August 2026. The date above is the public announcement date, not the date a repository was created or a third-party provider added the model. Where access or weights arrived later, that distinction is recorded below.
Release record
| Field | Verified detail |
|---|---|
| Announcement | 9 June 2026 |
| Availability or weight release | 9 June 2026 in Gemini Live API public preview and Google Translate, with Meet in private preview |
| Release type | hosted live speech-to-speech translation model |
| Access | Google API and product access; no released weights |
| Architecture | streaming audio model that detects source language and emits translated speech continuously |
| Maximum stated context | continuous session limits are documented in the Gemini Live API rather than as a launch token headline |
What changed
Gemini 3.5 Live Translate expanded speech translation from five Meet languages to more than 70 and more than 2,000 language combinations. The same named model shipped across developer, enterprise and consumer surfaces, with SynthID applied to generated audio. It therefore meets the archive threshold as a new audio foundation service.
The practical comparison is therefore not simply whether Gemini 3.5 Live Translate has the largest headline score. Teams need to ask whether its architecture, access terms, latency, tool behaviour and evaluation setup match the workload they actually intend to run. A model can lead one harness while losing on cost, refusal behaviour, multilingual quality or repeatability in another.
Benchmarks worth retaining
| Evaluation | Reported result | How to read it |
|---|---|---|
| Language coverage | 70+ automatically detected languages | Launch capability count, not a translation-quality score |
| Meet combinations | 2,000+ language combinations | Expanded from five supported languages in the previous experience |
| Streaming lag | a few seconds behind the speaker | First-party qualitative latency claim |
These are release-time results, not independently reproduced guarantees. The launch focuses on product coverage and qualitative latency rather than a complete public benchmark table; teams still need language-pair, accent, noise and domain-terminology evaluation. Scores should remain attached to the disclosed effort setting, agent harness, tool access, timeout, context-management policy and judge model. Moving a number into a procurement sheet without those conditions creates false comparability.
Architecture and access
Gemini 3.5 Live Translate is described as streaming audio model that detects source language and emits translated speech continuously with continuous session limits are documented in the Gemini Live API rather than as a launch token headline of stated context. Its access position at verification time is Google API and product access; no released weights. That wording matters: open weights, source-available weights, an API, a product preview and a research demonstration give adopters very different rights and different levels of reproducibility.
Before deployment, record the exact model identifier or checkpoint, inference stack, quantisation, reasoning setting, region, price schedule and supplier terms. If the release uses a custom licence, read the licence itself rather than relying on the word “open” in launch copy. If it is API-only, preserve the dated documentation and change-notice route because the served snapshot can change without a downloadable artefact.
What an evaluation should test next
For Gemini 3.5 Live Translate, a credible internal gate should include:
- a frozen set of representative tasks with pass, fail and abstain criteria;
- a matched baseline using the same tools, timeout, prompt budget and reviewer rubric;
- repeated runs to expose variance rather than reporting a single best attempt;
- latency, token use and total task cost alongside task success;
- adversarial, multilingual and long-context cases relevant to the real deployment; and
- rollback evidence showing the previous model can be restored safely.
The wider model change-control guide explains how to keep model, prompt, tool and corpus changes reconstructable. The AI dependency inventory guide covers the release and supplier records needed after deployment.
AIEngine verdict
This release belongs beside the other realtime voice models. Buyers should verify quality per language pair and measure lag, speaker preservation and failure recovery under real meeting conditions.
This is a launch assessment, not a certification. Benchmark leadership is useful evidence of where to test; it is not authorization to place the model in a high-impact workflow without domain evaluation, security review and an accountable owner.
Primary sources
- Google Gemini 3.5 Live Translate announcement
- Gemini Live Translate API documentation
- Gemini 3.5 audio model card
Image provenance
Hero image: Google official Gemini release artwork. The locally served WebP is a crop of the first-party release or model-card asset recorded in the repository provenance manifest.
