OpenAI: what the official pages do not settle
15 of 26 conditions we check across 2 listed plans are not stated on OpenAI's own pages, so a verdict on them is unknown. Each line says what one official sentence would have to state. Checked against the official pages as of 2026-11-11.
whisper-1, recorded audio (pay as you go)
OpenAI Speech to Text · 8 of 13 conditions not settled
- Real-time streaming: not stated. A sentence on OpenAI's official pages that says whether audio is handled in real time as it is made or arrives for this plan would change this verdict.
- Speaker diarization (an add-on is added to the cost): not stated. A sentence on OpenAI's official pages that says whether speakers are labelled, and any extra price for this plan would change this verdict.
- No minimum billed duration per request: not stated. A sentence on OpenAI's official pages that says whether audio is billed by the second with no minimum per request for this plan would change this verdict.
- Korean speech: not stated. A sentence on OpenAI's official pages that says whether Korean is supported for this plan would change this verdict.
- Recurring monthly free tier: not stated. A sentence on OpenAI's official pages that says whether the free allowance repeats every month for this plan would change this verdict.
- SOC 2 report: not stated. A sentence on OpenAI's official pages that says which SOC 2 report the provider holds for this plan would change this verdict.
- Data processing agreement: not stated. A sentence on OpenAI's official pages that says whether the vendor offers a data processing agreement for this plan would change this verdict.
- Service level agreement (SLA): not stated. A sentence on OpenAI's official pages that says whether the plan comes with a service level agreement for this plan would change this verdict.
Verdict for this plan at 100 hours of audio a month, as text.
gpt-live-transcribe, live audio (pay as you go)
OpenAI Speech to Text · 7 of 13 conditions not settled
- Recorded-audio (batch) transcription: not stated. A sentence on OpenAI's official pages that says whether recorded audio can be transcribed for this plan would change this verdict.
- No minimum billed duration per request: not stated. A sentence on OpenAI's official pages that says whether audio is billed by the second with no minimum per request for this plan would change this verdict.
- Korean speech: not stated. A sentence on OpenAI's official pages that says whether Korean is supported for this plan would change this verdict.
- Recurring monthly free tier: not stated. A sentence on OpenAI's official pages that says whether the free allowance repeats every month for this plan would change this verdict.
- SOC 2 report: not stated. A sentence on OpenAI's official pages that says which SOC 2 report the provider holds for this plan would change this verdict.
- Data processing agreement: not stated. A sentence on OpenAI's official pages that says whether the vendor offers a data processing agreement for this plan would change this verdict.
- Service level agreement (SLA): not stated. A sentence on OpenAI's official pages that says whether the plan comes with a service level agreement for this plan would change this verdict.
Verdict for this plan at 100 hours of audio a month, as text.
Send a correction
If one of these is already stated on a page we missed, or a sentence has changed, send the page address through https://github.com/scraping-api-plans/web-data-plan-finder/issues. Nothing is sent to OpenAI from this page.