GPT Transcribe and GPT Live Transcribe

2026-07-28 · Publisher : OpenAI · Domain : Artificial Intelligence

OpenAI releases GPT Transcribe for file transcription and GPT Live Transcribe for low-latency streaming transcription, with keyword and language hints.

OpenAI releases GPT Transcribe for file transcription and GPT Live Transcribe for low-latency streaming transcription, with keyword and language hints. ### What changes - GPT Transcribe: audio files, streamed transcriptions and final transcriptions of validated turns in a Realtime session. - GPT Live Transcribe: low-latency transcription deltas with adjustable latency. - Free-form context, keyword hints and several expected languages to better handle business terms and language switching. - Replace whisper-1 and gpt-4o-transcribe, whose end of service is set for February 26, 2027. ### Specifications - Audio input, text output. - Availability: Audio transcriptions API and Realtime API. - Context and knowledge cutoff: not disclosed. - Identifiers: gpt-transcribe, gpt-live-transcribe. ### Pricing | Model | Pricing | | --- | --- | | gpt-transcribe | $0.0045 per minute | | gpt-live-transcribe | $0.017 per minute | For comparison: gpt-4o-transcribe $0.006 per minute, gpt-4o-mini-transcribe $0.003 per minute. ### Benchmarks No figures published by the vendor in the announcement. ### Key takeaways The transcription lineup is being overhauled and the previous one is reaching end of life: integrations based on whisper-1 must migrate before February 2027.

Tags : Audio, OpenAI, Transcription, Parole vers texte, Realtime