Gemini 3.5 Live Translate is Google's audio model for live speech-to-speech translation. Google announced it in June 2026 with automatic detection of 70+ languages and said it is rolling out across three official surfaces: Google Translate, Google AI Studio / the Gemini Live API, and Google Meet speech translation preview. That means the practical setup depends on which surface you can access.
That's the important distinction. The Gemini app's Gemini Live voice assistant can translate prompts, but it is not the same documented surface as the Gemini 3.5 Live Translate model/API rollout. This guide walks through the official entry points, clears up the naming confusion, and shows when a browser-based real-time meeting translation tool is the better call.
A quick note on naming before we start. Google ships model updates fast, and the label you see in your app — Gemini in the 3.5 series, or whatever version has rolled out to your account — changes over time. The steps below work regardless of the exact version number on your screen.
Key Takeaways
- Official surfaces matter: Google named Google Translate, Google AI Studio / the Gemini Live API, and Google Meet speech translation preview for Gemini 3.5 Live Translate rollout.
- The Gemini app is different: Gemini Live can translate voice prompts, but do not assume it is the same 3.5 Live Translate model surface unless Google shows it in your account.
- Availability, languages, and the newest model depend on your plan, region, and product surface. Google updates all three often, so check Google's current pages before relying on a specific pair.
- For multilingual meetings, a browser tool such as MirrorCaption captures the meeting-tab audio with no bot joining the call and shows the original next to the translation.
What Is Gemini Live Translate?
Gemini 3.5 Live Translate is a Google model and rollout label, not a single app you download. Google describes it as a live speech-to-speech translation model reached by developers through the Gemini Live API, surfaced to consumers through Google Translate, and being introduced to Google Meet through speech translation preview.
The Gemini app still matters because people use "Gemini Live Translate" loosely. Gemini Live is the conversational voice mode inside the Gemini app, and it can translate if you ask it to. The safer way to describe it is this: Gemini Live may help with voice translation, while Gemini 3.5 Live Translate is the specific model/API and product rollout Google documented separately.
Think of it this way: Google Translate is the consumer translator surface, Google AI Studio / the Gemini Live API is the developer surface, and Google Meet preview is the meeting surface for eligible accounts. Which one you can use depends on where the conversation happens.
Illustrative example: Maria, a freelance designer in Lisbon, wants to test Google's live translation for a Portuguese conversation. If she is using the consumer route, she opens Google Translate and uses the live translation flow available in her app. If she is prototyping a product, she tests the Gemini Live API in Google AI Studio instead. Those are different surfaces with different limits.
How to Use Gemini 3.5 Live Translate, Step by Step
Start by choosing the surface that matches your use case. The same model family can show up in different Google products, but the setup is not identical.
Step 1: Choose Google Translate, Google AI Studio, or Google Meet preview
For consumer use, check the Google Translate app on Android or iOS. For developer testing, use Google AI Studio or the Gemini Live API documentation for gemini-3.5-live-translate-preview. For meetings inside Google Meet, check whether your Workspace account has access to Google's speech translation preview.
Step 2: Select or configure the target language
In Google Translate, select the language pair available in your app. In the Live API, configure the session and target language before streaming audio. In Google Meet preview, the meeting controls and account policy determine which translation options appear.
Step 3: Speak naturally and listen to the translation
Talk in your own language and let the active Google surface handle the translated output. Short, complete sentences still work better than long, rambling ones; the cleaner your input, the cleaner the translation.
Step 4: Switch to text or a transcript when you need a record
When you need something you can save, check what the surface actually provides. Google Translate, Google Meet preview, and a custom API app do not necessarily expose the same transcript, export, or retention controls. For sensitive conversations, review your account, workplace, and product settings before saving anything.
Want this for meetings instead?
If your real problem is a multilingual video call rather than a face-to-face chat, jump ahead to the meeting section — or just try MirrorCaption free and read live captions in your language while the call is still happening.
Google Translate vs. Gemini Live API vs. Google Meet Preview
This is where many people get confused, because Google offers several overlapping translation features under different names. Here's how the official Gemini 3.5 Live Translate surfaces differ.
| Tool | Best for | How it works | Catch |
|---|---|---|---|
| Google Translate app | Consumer live translation on Android or iOS | Use the translation app surface Google named for rollout | Availability and exact controls depend on app version, region, and account |
| Google AI Studio / Gemini Live API | Developer prototypes and custom apps | Stream audio to gemini-3.5-live-translate-preview | Requires code, API keys, billing controls, and product work |
| Google Meet speech translation preview | Meet calls where the preview is enabled | Native Google Meet meeting translation | Preview/rollout status and Workspace eligibility decide whether you see it |
Exact languages, model versions, and plan requirements change regularly. Check Google's Translate, AI Studio / Live API, and Meet documentation before you rely on a specific capability.
The pattern is clear once you see it. If you want the consumer translator, start with Google Translate. If you are building a product, use the Live API. If your call is in Google Meet and your account has preview access, use the native Meet option. If the call is on Zoom, Teams, Webex, or a client-controlled platform, you need a tool that works across meeting tabs.
Where Gemini Live Translate Works Well — and Where It Doesn't
Let's be fair about both sides. Gemini 3.5 Live Translate is genuinely impressive where Google exposes it, and genuinely constrained where your conversation happens outside those surfaces.
Where it shines
- Travel and errands: ordering food, asking directions, sorting out a hotel mix-up.
- One-on-one casual conversation where both people are patient and can take turns around a single phone.
- Learning and curiosity: translate a phrase, then ask why it's phrased that way — the follow-up is the magic.
- Quick personal help: reading a foreign-language menu, sign, or message out loud.
Where it gets frustrating
- Group conversations: one phone, one assistant, and three people talking at once is a recipe for crosstalk.
- Video meetings: Gemini listens through your phone's mic, so it can't cleanly hear remote participants coming through your laptop.
- Fast, overlapping speech: turn-taking with an assistant adds a beat of latency that breaks the rhythm of a real debate.
- A written record of who said what: Gemini is conversational, not built to label speakers across a long meeting.
Illustrative example: a five-person product standup spanning Berlin, São Paulo, and Seoul tries to run the whole call through one teammate's Gemini Live session. By minute three, people are talking over each other, the assistant keeps interrupting to confirm, and nobody can tell whose words just got translated. The tool isn't broken — it's just being asked to do a job it was never designed for.
For accuracy expectations across any AI translator, it helps to read up on how accurate real-time translation actually is before you lean on it for anything important. The honest summary: great for the gist, risky for the fine print.
Translating Meetings: Why a Browser Tool Fits Better
Here's the gap. A multilingual meeting happens on Zoom, Google Meet, Microsoft Teams, Webex, or whichever platform the host chose. Google is bringing Gemini 3.5 Live Translate into Google Meet speech translation preview, so some Meet calls may have a native option. For Zoom, Microsoft Teams, Webex, client-controlled platforms, or side-by-side original and translated captions, you still need something that works across any meeting tab.
This is the problem MirrorCaption was built for. It's a browser-based real-time transcription and translation tool that captures the audio from your meeting tab in desktop Chrome or Microsoft Edge — so no bot joins the call, and there's no install for the other participants. You read live captions in your language while the other person is still speaking, with the original shown side by side so you never lose the nuance.
| Need | Google Gemini 3.5 Live Translate surfaces | MirrorCaption |
|---|---|---|
| Translate a Zoom / Meet / Teams call | Meet preview only for eligible Google accounts; not a Zoom or Teams overlay | Yes — captures meeting-tab audio in Chrome or Edge |
| Bot joins the meeting? | No separate bot for Translate/API use; Meet behavior depends on native Meet controls | No bot — it stays outside the call |
| See original next to translation | Depends on the Google surface | Side-by-side original and translation |
| Speaker labels for who said what | Not built for it | Auto speaker detection |
| Speak the translation aloud | Supported through the live-audio model/API and selected product surfaces | Speak Translations voices your translated speech |
| Face-to-face on a phone | Google Translate app is the consumer path to check first | Talk mode — one continuous session, no push-to-talk |
The tools are built for different rooms. Use Google's own surfaces when your conversation fits Google Translate, the Live API, or eligible Meet preview. Use a browser translator when you're sitting in a video call with colleagues who don't share your first language and the host platform is outside your control. And for face-to-face conversation, MirrorCaption's Talk mode runs as one continuous session on your phone — you start it once and take turns naturally, rather than tapping for every sentence.
Illustrative example: after the messy standup, the same team switches approach. Each person opens MirrorCaption in their own browser, captures the meeting tab, and reads the call in their own language — German for the Berlin lead, Portuguese in São Paulo, Korean in Seoul. The crosstalk stops, because nobody is sharing one phone. Everyone follows along live, and the late joiner skims the running AI summary to catch up.
If you're weighing options for recurring multilingual calls, our roundup of the best meeting translator in 2026 compares the main tools head to head, and the Google Meet translation alternative page covers the Google-ecosystem angle specifically.
Pricing: What You Actually Pay
Gemini 3.5 Live Translate pricing depends on the surface. The Google Translate app is the consumer route to check on your phone. Developer access through the Gemini Live API has its own API pricing, limits, and preview terms. Google Meet speech translation depends on preview access and eligible Workspace availability. Because plan names and limits change, verify the current details in Google's own product pages before budgeting around it.
MirrorCaption keeps pricing simple. You start with 1 free hour to try, one-time, with no credit card and no monthly reset. The Pro Yearly plan is €54.99 and includes 100 hours of hosted transcription credit for the year. The Premium plan is a €99 one-time purchase that includes 200 hours of hosted transcription credit up front, all future product updates, and the lowest per-hour rate when you top up with Voice Packs (sold separately, from €2.99 for 5 hours). Premium is a one-time purchase, not a subscription — but it isn't unlimited hosted time; once the included hours run out, you add more with Voice Packs.
Final Recommendation
To use Gemini 3.5 Live Translate, start with the official surface that matches your situation: Google Translate app for consumer live translation, Google AI Studio / Gemini Live API for developer work, or Google Meet speech translation preview if your Workspace account has access. Do not assume the Gemini app, Pixel Live Translate, and Gemini 3.5 Live Translate are the same product surface.
When the conversation moves to a video call with colleagues across languages, reach for a browser-based tool instead. Read the call live, see the original beside the translation, let speaker labels keep track of who said what, and turn on spoken output when the other side needs to hear it. Different rooms, different tools.
The fastest way to feel the difference is to try it on your next call. It runs in your browser, so there's nothing for participants to install and no bot in the meeting.
Translate Your Next Meeting Live
1 free hour to try. No credit card. No monthly reset. No bot in your call.
Get Started FreeFrequently Asked Questions
Is Gemini 3.5 Live Translate free?
Consumer availability depends on the Google Translate app and your region. Developer access through the Gemini Live API has its own billing and preview limits, while Google Meet speech translation depends on preview access and eligible Workspace availability.
What languages does Gemini Live Translate support?
Google says Gemini 3.5 Live Translate detects 70+ languages, but exact language support depends on the surface: Google Translate app, Gemini Live API / Google AI Studio, or Google Meet preview. Check Google's current documentation before relying on a specific pair.
Can Gemini translate a Zoom or Google Meet meeting in real time?
Gemini 3.5 Live Translate is rolling out to Google Meet speech translation preview for eligible accounts, but it does not sit on top of Zoom, Teams, or Webex. To translate those meetings live, a browser-based tool such as MirrorCaption captures the meeting-tab audio without a bot joining the call.
Does Gemini Live Translate work without internet?
Gemini 3.5 Live Translate is a cloud/live-audio capability and needs an internet connection. Google Translate can download offline packs for phrase and text translation, but that is separate from the 3.5 Live Translate model.
How is Gemini Live Translate different from Google Translate?
Google Translate is the consumer translation app surface Google has named for Live Translate rollout. Gemini Live in the Gemini app is a voice assistant that can translate prompts, but it is not the same as the documented Gemini 3.5 Live Translate API/model surface.
Is Gemini Live Translate accurate enough for important conversations?
For travel, casual chat, and getting the gist, it's strong. For high-stakes settings such as legal or medical conversations, treat any AI translation as a draft and confirm critical details — ideally with a qualified human interpreter on record.