OpenAI
GPT Realtime Whisper
Measured 2026-09-21
Run GPT Realtime Whisper on Speko →One API, every model on this board.
WER · batch
—
Streaming-only endpoint
WER · streaming
Read speech
5.02%
Conversation
16.82%
Details
Read speech
n=1541 · FLEURS test+dev+train · vendor-direct
Conversation
14.0–19.9% · EdAcc n=283 · 60 speakers, 2 passes, 2026-10-01
Finalize
1.14s
p50–p90 · US East · n=200 · vendor-direct · ±8%
Commit-to-final: the probe commits at speech end, so the vendor never decides the turn. The two US sweeps disagreed beyond the artifact’s bound; Singapore reads 1.34s.
Time to first token
1.81s
p50–p90 · US East · n=200 · vendor-direct · ±5%
Cost / min
$0.017
Billed by audio duration at $0.017/minute, not by token (developers.openai.com model page, read 2026-09-21). Same rate as gpt-live-transcribe.
Share this result
Embed the live badge
[](https://benchmarks.speko.ai/stt/openai-gpt-realtime-whisper)
<a href="https://benchmarks.speko.ai/stt/openai-gpt-realtime-whisper"><img src="https://benchmarks.speko.ai/badge/stt/openai-gpt-realtime-whisper.svg" alt="Speko stt rank"></a>
https://benchmarks.speko.ai/stt/openai-gpt-realtime-whisper