OpenAI's transcription models removed on 26 February 2027, and what to use instead
Last updated:
⏳ 144 days left — the old API shuts down February 26, 2027.
On August 26, 2026, OpenAI told developers that whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe and gpt-4o-transcribe-diarize will all be removed from the API on February 26, 2027, and recommended gpt-transcribe or gpt-live-transcribe instead (OpenAI deprecations, "2026-08-26: Transcription models"). Plain transcription can usually move to gpt-transcribe by changing the model name. Apps that rely on whisper-1-only features (SRT/VTT subtitles, word timestamps, /v1/audio/translations) need either a rewrite or a whisper-1-compatible endpoint.
What is removed on February 26, 2027
| Model | Removed | OpenAI's recommended replacement |
|---|---|---|
whisper-1 | February 26, 2027 | gpt-transcribe or gpt-live-transcribe |
gpt-4o-transcribe | February 26, 2027 | gpt-transcribe or gpt-live-transcribe |
gpt-4o-mini-transcribe | February 26, 2027 | gpt-transcribe or gpt-live-transcribe |
gpt-4o-transcribe-diarize | February 26, 2027 | gpt-transcribe or gpt-live-transcribe |
Related: OpenAI's deprecations page also lists tts-1 and tts-1-hd (text-to-speech) for removal on January 6, 2027, with gpt-realtime-2.1-mini recommended.
Which replacement fits your use
| You use the model for… | Simplest path | What to check |
|---|---|---|
Plain text transcripts (json / text) | Change model to gpt-transcribe | Accuracy and cost on your audio |
Subtitles (srt, vtt) or word/segment timestamps | A whisper-1-compatible endpoint, such as the whisper-1 drop-in on Apify | OpenAI's speech-to-text guide documents timestamp_granularities[] only for whisper-1 |
Translation to English (/v1/audio/translations) | A whisper-1-compatible endpoint | OpenAI's guide documents /v1/audio/translations only with whisper-1 |
| Live captions | gpt-live-transcribe | Realtime API usage differs from file upload |
Speaker labels (diarized_json) | Re-test on gpt-transcribe, or a separate diarization step | Output shape changes |
Prices
| Model | Price (OpenAI pricing page, October 2026) | Status |
|---|---|---|
whisper-1 (listed as "Whisper") | $0.006 / minute | Removed Feb 26, 2027 |
gpt-4o-transcribe | token-based, estimated $0.006 / minute | Removed Feb 26, 2027 |
gpt-4o-mini-transcribe | token-based, estimated $0.003 / minute | Removed Feb 26, 2027 |
gpt-transcribe | $0.0045 / minute | Recommended |
gpt-live-transcribe | $0.017 / minute | Recommended (realtime) |
| whisper-1 drop-in on Apify | $0.006 / minute, billed per started 15 s | Independent; Whisper small model |
Source: OpenAI pricing.
Keeping whisper-1's API: the two-line change
If you need whisper-1's exact request and response formats after February 26, 2027, keep model="whisper-1" and point the official SDK at a compatible endpoint:
- client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])
+ client = OpenAI(base_url="https://dropin-apis--whisper-compat.apify.actor/v1", api_key=os.environ["APIFY_TOKEN"])The full guide, with Node.js and curl examples, real SRT output and limits, is on the whisper-1 replacement page.
FAQ
Is gpt-4o-transcribe being deprecated?
Yes. OpenAI's August 26, 2026 notice removes gpt-4o-transcribe, together with gpt-4o-mini-transcribe, gpt-4o-transcribe-diarize and whisper-1, from the API on February 26, 2027.
What replaces gpt-4o-transcribe and whisper-1?
OpenAI recommends gpt-transcribe, or gpt-live-transcribe for realtime audio.
When exactly are OpenAI's old transcription models removed?
February 26, 2027, for all four models.
Does gpt-transcribe support SRT, VTT and word timestamps?
OpenAI's speech-to-text guide documents timestamp_granularities[] and the translations endpoint only for whisper-1. Test the subtitle formats you need on gpt-transcribe before migrating, or use a whisper-1-compatible endpoint.
Is gpt-transcribe cheaper than whisper-1?
Yes, on OpenAI's October 2026 price list: $0.0045 per minute versus $0.006 per minute for whisper-1.