GPT Transcribe

Audio modelCard summarized

High-accuracy speech-to-text model for file and Realtime input transcription

● Re-checked 12 Oct 2026 · 46 facts · 1 correctionSource: GPT Transcribe model page and transcription guides (OpenAI API docs); no system card published

Use it via API

$0.0045 / min

Code example Compare

Where to call it

OpenAI APIgpt-transcribe

Supports

Streaming

Full feature support

Good for

Transcribing completed recordings in their original language
Domain terms, multilingual audio and code-switching via prompt, keywords and languages
Detected-language output

Watch out for

No speaker labels, timestamps, subtitles or translation
Keywords can leak into the transcript
One bad hint rejects the request

Overview

OpenAI · 01Transcribe

GPT Transcribe

High-accuracy speech-to-text model for file and Realtime input transcription

Typical price

$0.0045 / min

Transcription

Max length: —
Languages: —
Modes: speech-to-text, streaming
Released: 28 Jul 2026

OpenAI published no system card for GPT Transcribe. Released 28 July 2026, it is OpenAI's recommended starting model for transcribing recorded speech in its original language: completed files, streamed file transcripts and committed turns in Realtime sessions. It accepts free-form context, keyword hints and several expected languages, and returns the languages it detects. model page; changelog; transcription guide

gpt-transcribe

Typical price: $0.0045 / min
Weights: Closed
Released: 28 Jul 2026

speech-to-text
streaming
realtime transcription (committed turns)
context prompt and keyword hints
multiple language hints / code-switching