LiveLingoQuick Answer: Can GPT-Live do live translation?
Yes. GPT-Live performs live translation by prompt, and it is included in every ChatGPT plan as of July 8, 2026. Open ChatGPT on iOS, Android, or the web, start a Voice session, and say something like "simultaneously translate whatever I say into Spanish." Because GPT-Live-1 and GPT-Live-1 mini are full-duplex (they listen and speak at the same time), the translation streams while you are still talking instead of waiting for you to finish. Free accounts get GPT-Live-1 mini; paid Go, Plus, and Pro plans get the larger GPT-Live-1. There is no dedicated translator interface (no language-pair picker, no side-by-side transcript), no published language list, and no API at launch. Developers who need OpenAI live translation programmatically use the separate gpt-realtime-translate model ($0.034/min, 70+ input languages, 13 output languages).
1. What OpenAI Shipped on July 8
OpenAI announced GPT-Live on Wednesday, July 8, 2026 in a blog post and livestream (Introducing GPT-Live). Two models shipped: GPT-Live-1, the new default voice model for paid Go, Plus, and Pro plans, and GPT-Live-1 mini, the new default for free accounts (per MacRumors and The Decoder). Both replace Advanced Voice Mode as the ChatGPT Voice default, with the old mode remaining selectable in voice settings.
The rollout covers iOS, Android, and chatgpt.com over the days following launch. The desktop app's voice mode is not included at launch, and ChatGPT Business, Enterprise, and Edu workspaces are excluded for now (per 9to5Mac and WinBuzzer). No new pricing was introduced; GPT-Live is included in existing plans. OpenAI cites more than 150 million people talking to ChatGPT through features like Voice and Dictation (TechCrunch).
The defining feature is full-duplex audio: the model processes what it hears while it is speaking, making interaction decisions many times per second, so it can be interrupted naturally, respond with backchannels ("mhmm," "got it") while you talk, or stay silent while you think (MarkTechPost). For anything requiring web search, deeper reasoning, or agentic work, GPT-Live delegates to GPT-5.5 in the background and keeps the conversation going while the result comes back. Users can pick three reasoning levels (Instant, Medium, High); the higher levels route to GPT-5.5 Thinking.
OpenAI's published evaluations focus on capability rather than speed: at High reasoning, GPT-Live-1 scores 84.2% on GPQA versus 45.3% for Advanced Voice Mode, and 75.2% on BrowseComp versus 0.7%. In human preference testing, GPT-Live-1 was preferred over Advanced Voice Mode in 75.7% of conversations, and mini in 69.2% (The Decoder, citing OpenAI's announcement). Notably, OpenAI published no latency numbers for GPT-Live at launch.
2. How Live Translation Works on GPT-Live
Live translation is one of the capabilities OpenAI explicitly names in the launch materials: the full-duplex architecture lets the model "engage in more natural back-and-forth, maintain a better sense of time, and even perform live translation" (OpenAI announcement). In press briefings, OpenAI demoed the model speaking a running translation while the presenter talked.
It is a prompt, not a mode
There is no translator toggle, language-pair picker, or dedicated translation UI. You start a Voice session and instruct the assistant: "I'd like you to simultaneously translate whatever I'm saying into Hindi." The model then renders a spoken translation as you speak, and continues until you tell it to stop or switch. This is the same prompt-invoked pattern ChatGPT Voice has had since the May 2026 Realtime release; what changed is the full-duplex delivery, which lets the translation overlap your speech instead of waiting for your turn to end.
What the launch demo showed
The only public translation-quality datapoint from launch week comes from OpenAI's own press demo. TechCrunch, watching a live English-to-Hindi translation, reported that "the assistant had a heavy American accent and spoke in Hindi that was unnatural sounding and had slightly bookish tone" (TechCrunch). That matches OpenAI's own disclosure: voice is optimized for "most spoken languages," and "for certain languages, the model may have a non-native accent or gaps in fluency." OpenAI has not published a supported-language list for GPT-Live, and full multilingual parity is explicitly listed as not available at launch (MarkTechPost).
The assistant-first trade-off
GPT-Live is a general conversational assistant that translates on request, not a dedicated interpreter. That shape has consequences for translation work. The backchannel behavior that makes conversation feel natural ("mhmm," "yeah") injects non-translation audio into a translation session, and day-one users widely reported the backchannels as intrusive (BigGo). Long pauses or background speech can trigger unintended responses, and reviewers with preview access reported the model interrupting and laughing mid-sentence before OpenAI tuned the behavior down (Simon Willison). A model whose defining skill is deciding when to speak is making that decision constantly during translation, where the correct answer is almost always "keep translating, say nothing else."
3. GPT-Live vs OpenAI's Purpose-Built Translator
The launch coverage largely missed that OpenAI already ships a dedicated live-translation model, and it is a different stack. On May 7, 2026, OpenAI released gpt-realtime-translate in the Realtime API: a purpose-built streaming speech-to-speech translation model trained on thousands of hours of professional interpreter audio, configured to remain translation-only and wait for enough context before producing speech (OpenAI announcement). GPT-Live and the Realtime models are separate families: GPT-Live is the ChatGPT consumer surface with no API, and gpt-realtime-translate is the developer surface with published specs, pricing, and constraints.
| GPT-Live (ChatGPT) | gpt-realtime-translate (API) | |
|---|---|---|
| What it is | Full-duplex voice assistant; translation by prompt | Dedicated streaming speech-to-speech translation model |
| Released | July 8, 2026 | May 7, 2026 |
| Access | ChatGPT app (iOS, Android, web); no API | Realtime API, endpoint /v1/realtime/translations |
| Price | Included in ChatGPT plans (free tier gets mini) | $0.034 per minute of audio |
| Languages | Unpublished; "most spoken languages," accent gaps disclosed | 70+ input languages; 13 output languages (documented) |
| Stays in translator role | No guarantee; conversational assistant, backchannels | Yes; translation-only by design, no system prompts |
| Transcripts | Chat transcript of the session | Text deltas of source and translated speech |
| Voice | 9 remastered ChatGPT voices | Dynamic voice adapting to source speaker's tone |
| Independent measurement | Not possible yet (no API) | Measured; see Section 5 |
The 13 documented output languages of gpt-realtime-translate: English, Spanish, Portuguese, French, German, Italian, Russian, Chinese, Japanese, Korean, Hindi, Indonesian, and Vietnamese (OpenAI Cookbook). Until OpenAI publishes a GPT-Live language list, this is the best available signal of where OpenAI's speech-translation training is strongest.
Two launch-week corrections worth making explicit, because early coverage garbled both: GPT-Live has no API pricing (the "$32 per million audio input tokens, $64 output" figures circulating in some posts belong to gpt-realtime-2, a different model), and GPT-Live is not available via the API at all; OpenAI's wording is that the models come to the API "soon," with a notification signup form and no timeline (Apidog's API-status roundup).
4. Limits OpenAI and Launch Reviewers Disclose
- Usage caps are real but unpublished. OpenAI's help center states voice usage limits vary by plan and Voice option and may change, with an in-app notice as you approach the limit. Third-party reports of specific numbers conflict (from 15 minutes daily to 2 hours daily for free accounts); treat any specific figure as unverified.
- No published latency numbers. OpenAI released capability benchmarks (GPQA, BrowseComp, preference rates) but no first-audio latency or lag measurements for GPT-Live.
- Language coverage is undocumented. "Most spoken languages," non-native accents and fluency gaps disclosed for others, full multilingual parity explicitly not at launch.
- Backchannels cannot be fully disabled. You can instruct the model to stay quiet, but day-one reports describe the "mhmm" behavior as persistent and, for many users, irritating.
- No video or screen sharing at launch, and the mode is reported unavailable in some contexts (Business/Enterprise/Edu workspaces; desktop app voice).
- Safety monitoring can interject. Per the GPT-Live system card, real-time monitoring can steer or interrupt a response, play a spoken safety message, or end high-risk conversations.
5. What Independent Measurement Shows for OpenAI Translation
What we measured (and what we did not)
GPT-Live cannot be measured programmatically yet: it has no API, and its consumer surface adds client-side voice detection and app behavior no harness can isolate. The numbers below are for OpenAI's API-tier translation model, gpt-realtime-translate, measured June 10, 2026 with the same energy-VAD utterance boundaries applied uniformly to every API-tier system in the LiveLingo benchmark. They are the best available floor for what OpenAI's speech-translation stack does on real audio. They are not GPT-Live numbers, and we label them accordingly. When GPT-Live gets an API, we will run the same protocol against it and update this section.
Reproducibility
Every number reproduces from the same three 120-second VOA public-domain audio clips, the same Realtime API endpoint, and the same Python harness used for the original four-system benchmark. The audio (audio.zip), raw per-utterance JSON (openai-realtime-results.json), and methodology are published at livelingo.io/research/benchmark-2026.
| System | Comprehension (0–5) | Utterance-end → translation arrival | Speed to first output (audio-to-audio only) | Output surface |
|---|---|---|---|---|
| LiveLingo | 4.96 | 1,518 ms | – | Streaming text + audio |
| Gemini 3.5 Live Translate | 4.93 | ~3,100 ms (drifts up to 13.9 s) | 2,947 ms | Audio (text sidecar) |
| Google Cloud STT v2 + Translate v3 | 4.77 | ~26,736 ms | – | Transcript |
| Azure Speech Translation | 4.65 | ~4,755 ms | – | Transcript |
| Whisper + GPT-4o-mini (DIY) | 4.63 | 2,720 ms | – | Transcript |
| OpenAI gpt-realtime-translate | 4.53 | ~3,800 ms (drifts up to 20.3 s) | 711 ms | Audio + transcript |
Three findings from that measurement frame what to expect from GPT-Live translation, because both products draw on the same speech-translation research direction at OpenAI:
OpenAI's stack is the fastest to first audio. Median 711 ms from start of speech to first translated audio, roughly four times faster than Gemini 3.5 Live Translate on the same metric. Full-duplex GPT-Live should inherit this responsiveness; it is the architecture's strength.
Continuous speech accumulates lag. Translated speech is bounded by the speaking rate of the synthesized voice, so on dense continuous audio the translation fell progressively behind, median 3.8 s and drifting to 20.3 s. A full-duplex model does not remove this constraint: spoken output still cannot catch up faster than human pace.
Comprehension trails purpose-built rivals. 4.53/5 composite, the lowest of the six systems measured, with recurring error classes of extraneous insertions, meaning inversions, and proper-name substitutions. The press-demo Hindi critique of GPT-Live (American accent, bookish tone) is consistent with a stack whose speed leads and whose fidelity trails.
6. How to Use GPT-Live for Translation Today
- Update the ChatGPT app on iOS or Android (or use chatgpt.com). GPT-Live is rolling out over the days following July 8; if your voice mode still behaves turn-based, the rollout has not reached your account yet.
- Tap the Voice icon in the message composer to start a session. Free accounts run GPT-Live-1 mini; Go, Plus, and Pro accounts run GPT-Live-1.
- Give the translation instruction up front: "Simultaneously translate everything I say into Japanese. Do not answer questions, do not comment, only translate." The explicit constraint reduces, but does not eliminate, conversational interjections.
- For two-way conversation, ask for both directions: "Translate my English into Japanese, and translate any Japanese you hear into English."
- Expect usage limits: OpenAI caps voice by plan without publishing numbers, and an in-app notice appears as you approach the cap.
7. When GPT-Live Fits, and When It Does Not
GPT-Live is the right choice when
- You want free, zero-install live translation and already use ChatGPT. The free tier now includes a full-duplex voice model, which no other major assistant offered before July 8.
- The conversation is casual and short: asking directions, a quick exchange, travel small talk. Backchannel quirks and role-drift matter less.
- You value the assistant around the translation: you can ask follow-up questions, get context, or have GPT-5.5 look something up mid-conversation.
A different tool fits better when
- You need the translator to stay in role for a sustained conversation. A dedicated interpreter surface does not backchannel, laugh, or answer instead of translating.
- You need a readable two-language transcript while listening. GPT-Live gives you a chat log, not a side-by-side source-and-translation view that never rewrites itself.
- You need documented language coverage. OpenAI has not published what GPT-Live supports; purpose-built products document their pairs.
- You need translated phone calls. None of the OpenAI surfaces dial a phone number. LiveLingo places outbound calls to regular phone numbers with translation running on the line, where the recipient needs no app.
An honest concession. LiveLingo (publishing this guide) sits in the dedicated-translator category: streaming text plus audio, per-speaker attribution, displayed transcripts that never retract, and translated outbound phone calls. GPT-Live's conversational naturalness is a real advance, its full-duplex turn-taking is ahead of every dedicated translator app including ours, and its free tier makes it the easiest way for anyone to try live voice translation today. Side-by-side specs: /compare/chatgpt-translation. Benchmark numbers: /research/benchmark-2026. OpenAI's full translation lineup: /guides/openai-live-translation.
8. Frequently Asked Questions
What is GPT-Live?
GPT-Live is OpenAI's full-duplex voice model family for ChatGPT, debuted July 8, 2026. Two models shipped: GPT-Live-1 (default for paid Go, Plus, and Pro plans) and GPT-Live-1 mini (default for free users). Unlike the Advanced Voice Mode it replaces, GPT-Live listens and speaks at the same time, so it can be interrupted naturally, backchannel while you talk, and perform live translation while you are still speaking. It delegates complex queries to GPT-5.5 in the background. It runs on iOS, Android, and chatgpt.com; there is no API at launch.
Can GPT-Live translate in real time?
Yes. Live translation is an explicitly named launch capability, invoked by prompt rather than a mode toggle: tell the assistant to "simultaneously translate whatever I say into Japanese" and it speaks a running translation while you talk. OpenAI says voice is optimized for "most spoken languages" without publishing a list, and discloses that some languages may get a non-native accent or gaps in fluency. In OpenAI's own press demo, TechCrunch observed the live Hindi translation had a heavy American accent and an unnatural, slightly bookish tone.
Is GPT-Live free?
Yes, with a tier split. GPT-Live-1 mini is the default voice model for free ChatGPT accounts; paid Go, Plus, and Pro plans get the larger GPT-Live-1. No new subscription was introduced. Voice usage limits vary by plan and may change; OpenAI publishes no fixed minute caps, and third-party reports of specific daily limits conflict with each other.
Does GPT-Live have an API?
No. At launch GPT-Live is ChatGPT-only. OpenAI plans to bring the models to the API "soon" and offers a notification signup form, with no endpoints, model IDs, pricing, or timeline published. Developers who need OpenAI live translation today use the separate Realtime API model gpt-realtime-translate: $0.034 per minute, 70+ input languages, 13 output languages.
How is GPT-Live different from Advanced Voice Mode?
Advanced Voice Mode was speech-to-speech but turn-based: it waited for silence to decide you were done, then replied. GPT-Live is a single full-duplex model that processes incoming audio while generating output, making interaction decisions many times per second. It can be interrupted mid-sentence, backchannels while you speak, stays quiet while you think, and hands complex questions to GPT-5.5 in the background. In OpenAI's human preference testing, GPT-Live-1 was preferred over Advanced Voice Mode in 75.7% of conversations. Advanced Voice Mode remains selectable in settings.
What languages does GPT-Live support for translation?
OpenAI has not published a language list for GPT-Live. Launch materials say voice is optimized for "most spoken languages" and disclose accent and fluency gaps for others; full multilingual parity is explicitly not available at launch. OpenAI's developer-facing translation model, gpt-realtime-translate, documents 70+ input languages and 13 output languages: English, Spanish, Portuguese, French, German, Italian, Russian, Chinese, Japanese, Korean, Hindi, Indonesian, and Vietnamese.
How does GPT-Live compare to Gemini Live for translation?
Different architectural bets. GPT-Live is a full-duplex conversational assistant that translates on request. Google's Gemini 3.5 Live Translate is a dedicated streaming speech-to-speech translation model in the Google Translate app, Google Meet, and the Gemini Live API, auto-detecting 70+ languages and preserving speaker prosody. Gemini's surface is measurable through a public API, where LiveLingo Research recorded a 4.93/5 comprehension composite and a roughly 3-second speaking delay. GPT-Live has no API, so no equivalent independent measurement exists yet; the launch-week Hindi demo critique is the only public quality datapoint.
How do you switch back to Advanced Voice Mode from GPT-Live?
Open ChatGPT voice settings and select Advanced Voice Mode. GPT-Live replaced it as the default on July 8, 2026, but OpenAI kept the old mode selectable for users who need features GPT-Live does not support at launch, such as video and screen sharing.
How do you stop GPT-Live from interrupting or saying "mhmm"?
You cannot fully disable the backchannel behavior at launch. The workaround is an explicit instruction at the start of the session, such as "stay silent until I address you directly" or, for translation, "only translate what I say, add nothing else." Day-one reports say this reduces but does not eliminate interjections, and long pauses or background speech can still trigger unintended responses. If the behavior blocks your use case, Advanced Voice Mode remains selectable in voice settings.
Which tool should you use for live conversation translation right now?
GPT-Live fits casual, hands-free translation inside an assistant you already use, on a free tier. Purpose-built translators fit sustained real conversations better: they hold a translation-only role, show a readable two-language transcript, and are measured on fidelity. On the LiveLingo 2026 benchmark, OpenAI's measurable translation surface (gpt-realtime-translate) scored 4.53/5 comprehension with translated audio drifting up to 20.3 seconds behind on continuous speech, while transcript-first systems held higher fidelity with bounded delay. For translated calls to regular phone numbers, none of the OpenAI surfaces dial; that is a different product category.
9. Sources
- OpenAI. Introducing GPT-Live. OpenAI blog, July 8, 2026. openai.com
- OpenAI. GPT-Live system card. OpenAI Deployment Safety, July 2026. deploymentsafety.openai.com
- OpenAI. ChatGPT Voice (help center). help.openai.com
- TechCrunch. OpenAI releases new voice models for more natural live conversations, July 8, 2026. techcrunch.com
- MarkTechPost. OpenAI Releases GPT-Live and GPT-Live-1 mini: Full-Duplex Voice Models That Delegate Deeper Reasoning to GPT-5.5, July 8, 2026. marktechpost.com
- MacRumors. OpenAI Introduces GPT-Live to Make ChatGPT Voice Feel Like a Real Conversation, July 8, 2026. macrumors.com
- The Decoder. ChatGPT can now listen and talk at the same time, July 2026. the-decoder.com
- SiliconANGLE. OpenAI launches GPT-Live voice model series ahead of broad GPT-5.6 release, July 8, 2026. siliconangle.com
- 9to5Mac. OpenAI upgrading ChatGPT with all-new voice mode experience, July 8, 2026. 9to5mac.com
- Simon Willison. Introducing GPT-Live (notes from preview access), July 8, 2026. simonwillison.net
- OpenAI Developers. gpt-realtime-translate model reference and realtime translation guide. developers.openai.com
- Apidog. Is There a GPT-Live API? What Developers Can Use Instead, July 2026. apidog.com
- LiveLingo Research. Real-Time Voice Translation Benchmark 2026, OpenAI gpt-realtime-translate addendum, June 10, 2026. livelingo.io/research/benchmark-2026. Dataset DOI: 10.5281/zenodo.21250032
- LiveLingo. ChatGPT Translation in 2026: Voice Mode, gpt-realtime-translate API, and Whisper Compared. livelingo.io/guides/openai-live-translation
Release date, model names, tier mapping, rollout scope, and evaluation figures verified against OpenAI's announcement, system card, help center, and the launch coverage linked above on July 9, 2026. GPT-Live is one day old at publication: OpenAI may change tiers, limits, language coverage, and model behavior quickly; consult the linked sources for current state before relying on any specific claim.