Sauti / openai
OpenAI whisper-1
Prior generation, still offered and still priced
WER
13.5%
95% CI 11.6%–15.8%
Rank
#9
of 12 measured
Latency
2.9s
mean, per clip
Cost
$0.36
per audio hour
10 other models are a statistical tie with this one — their confidence intervals overlap, so the ordering between them is not something this data can settle: nova-3, gpt-transcribe, Whisper large-v3 (self-hosted), universal-2, universal-3-5-pro, Whisper small (self-hosted), gpt-4o-mini-transcribe, scribe_v2, scribe_v1, gpt-4o-transcribe.
What it is
OpenAI's hosted Whisper endpoint. Absent from their main pricing table but still documented and priced on its own model page, and not marked deprecated.
Pick it when
Only for continuity with an existing integration. It is on the board to measure the generational delta, not because we would recommend starting here.
Think twice when
On price alone: it is the joint most expensive OpenAI option and gpt-transcribe placed ahead of it for less. If you want Whisper specifically, self-hosting the same architecture placed better here at no marginal cost, though far slower on our CPU bridge.
Assessment written against the board of 2026-08-08. The figures above are read live from the current published snapshot (2026-08-08), so if those dates differ, trust the figures.
What the vendor publishes
openai publishes no stated word error rate figure for this model on any public benchmark. That absence is itself the finding. Source (read 2026-08-15).No accuracy figure published. OpenAI also does not document WHICH Whisper artifact whisper-1 serves, so the paper's per-size numbers cannot honestly be attributed to this API model.
How this was measured
250 clips (2.9 hours) of unscripted conversational English from People’s Speech (MLCommons, CC-BY), human-transcribed. Scored on a held-out split (N=77) never used to tune anything, through one fixed text normalizer, with bootstrap 95% confidence intervals. One corpus, one language, one modality: batch pre-recorded. It does not tell you how this model handles your accents, your domain vocabulary, or streaming.
Cost is openai’s own published pre-recorded pay-as-you-go list price, read from their pricing page on 2026-08-08. Streaming and committed-volume rates differ.Prior generation, still offered. Same rate as gpt-4o-transcribe, which is why the newer gpt-transcribe at $0.0045/min supersedes it on price as well as accuracy.