Model · OpenAI · Speech-To-Text
GPT Live Transcribe
OpenAI's realtime transcription model — the streaming half of the pair replacing GPT-4o Transcribe. Built for live captioning and voice agents, with tunable latency and keyword hints.
- Modality
- Speech-To-Text
- License
- Proprietary (Proprietary)
- Released
- July 28, 2026
- Last verified
- September 8, 2026
- Runs locally
- No
- Also handles
- realtime-voice
Strengths
- Tunable latency, so you can trade delay against accuracy per use case
- Keyword hints and multiple language hints improve accuracy on domain vocabulary
- Accepts unstructured context to prime the transcript
Weaknesses
- Realtime transcription sessions endpoint only — no batch file transcription
- Roughly 3.8x the per-minute cost of GPT Transcribe ($0.017/min vs $0.0045/min)
- OpenAI does not publish a context window or max output for this model
- Closed weights — no self-hosting
Try it
| Where | Type | Notes |
|---|---|---|
| OpenAI API | hosted-api | API key required |
Version history
- GPT Live Transcribe Jul 2026 Current
- GPT Transcribe Jul 2026
Official sources
- Model docs docs
- API deprecations docs
Change log
- — Initial entry. Released 2026-07-28. Named alongside gpt-transcribe on OpenAI's deprecations page as a replacement for gpt-4o-transcribe (shutdown 2027-02-26). Context window and max output are not published by OpenAI.
Esc