CekuraBench

Zoom Scribe Pro

Every modelZoom

Accuracy vs. speed

Lower and further left is better

Best quadrantBest picks
Final-text delay P50 (ms)
Pipecat Dataset WER (%)

Zoom Scribe Pro is 3rd of 19 on the Pipecat Dataset and 16th of 19 on the Ocular Dataset.

1.96% Pipecat WER (3rd) · 4.84% Ocular WER (16th) · 1.70s First text (13th) · 162ms Final text (10th)Likely range, 95% confidence: Pipecat WER 1.73 to 2.20%, Ocular WER 4.23 to 5.43%Published price $0.25 per hour of streamed audio. Scribe API Live, paid from prepaid credits. Zoom publishes one Live rate and does not list Pro separately. Source, checked Oct 8, 2026.

Last updated Methodology by Dileep Chagam

Zoom Scribe Pro against every other model

Lower is better on every measure. Hover a column for that model's full numbers.

  1. 1.8%
    Universal 3.6 Pro
  2. 1.9%
    Universal 3.5 Pro
  3. 2.0%
    Scribe Pro
  4. 2.0%
    Scribe
  5. 2.0%
    Chirp 3
  6. 2.1%
    Reson8
  7. 2.2%
    Realtime Whisper
  8. 2.4%
    Ink 2
  9. 2.5%
    Linden
  10. 2.5%
    STT-RT v5
  11. 2.6%
    Chirp 2
  12. 2.7%
    STT-1
  13. 3.4%
    4o Mini Transcribe
  14. 3.5%
    Nova-3
  15. 3.7%
    Pulse
  16. 3.9%
    4o Transcribe
  17. 3.9%
    Flux English
  18. 5.0%
    Flux Multilingual
  19. 6.5%
    Gradium
Glossary: WER, TTFT, TTFS, P50
WER · Word error rate
Share of words the model got wrong: replaced, missing or added. Lower is better.
Pipecat Dataset
1,000 short public clips from Pipecat's STT benchmark dataset.
Ocular Dataset
8 recordings from 4 real phone conversations, licensed from Ocular.
Likely range · 95% range
Where the model's true score probably falls, given how much audio was tested. Overlapping ranges are too close to call.
TTFT · Time to first text
How long after the audio starts streaming until the first words appear.
TTFS · Final-text delay
How long after the caller stops until the full transcript arrives. A voice agent waits for this before it replies.
P50 / P90 / P95 · Typical, slow and slowest turns
Half of turns are faster than P50. Only 1 in 10 is slower than P90, and 1 in 20 slower than P95.

Where Zoom Scribe Pro struggled

On real calls, dropped words are its biggest source of error. On clean clips, wrong words lead.

Dropped words

What the speaker said, with Zoom Scribe Pro's mistakes marked: The audiobook I am currently listening to has a narrator who speaks much (left out) too fast and I'm trying to find the setting in the app that will let me slow the playback speed and

Pro also ends on an “Okay” that is not in the reference and that no other model wrote.

Everything Zoom Scribe Pro wrote

The audiobook I'm currently listening to has a narrator who too fast, and I'm trying to find the setting in the app that will let me slow the playback speed, and… Okay.

Pipecat Dataset clip pipecat-e946b3da-966a-dcc7-2a7c-fc4e9d987a62

Wrong name

What the speaker said, with Zoom Scribe Pro's mistakes marked: I'm looking for a book I read a long time ago about a journeyer to the center of the earth but it is not the famous one by Jules Verne written as Julius Venu it is

Everything Zoom Scribe Pro wrote

I'm looking for a book I read a long time ago about a journey to the center of the Earth, but it is not the famous one by Julius Venu, it is…

Pipecat Dataset clip pipecat-1d1b8787-f156-3be6-d1b1-40d1f2663783

Marked: words Zoom Scribe Pro left out or got wrong. Audio from the Pipecat Dataset, under CC BY 4.0, re-encoded as MP3.

TTFT and TTFS at P50, P90 and P95

Timed on the same 206 streamed Ocular turns as every other model. P90 and P95 show what a caller meets on a slow turn.

TTFS, final text

How long after the caller stops until the full transcript arrives. A voice agent waits for this before it replies.

20ms100ms500ms2s
Typical, P50
162ms
10th fastest of 19
Slow, P90
387ms
13th fastest of 19
Slowest, P95
541ms
12th fastest of 19

TTFT, first text

How long after the audio starts streaming until the first words appear.

200ms1s5s20s
Typical, P50
1.70s
13th fastest of 17
Slow, P90
9.13s
Slowest of 17
Slowest, P95
15.10s
Slowest of 17

Every other model's typical turn, on the same axis.

1,000 Pipecat clips · 8 Ocular recordings · 206 turns timed · 0 failed requests

Zoom Scribe Pro head to head

Every comparisonAll 19 modelsBenchmark source

Zoom Scribe Pro FAQ

  • Zoom Scribe Pro places highest on the Pipecat Dataset, 3rd of 19 at 1.96%. Its weakest placing is the Ocular Dataset, 16th of 19 at 4.84%.

  • Zoom Scribe Pro transcribes the 1,000 public Pipecat clips at 1.96% word error rate and the eight licensed Ocular recordings at 4.84%. Word error rate counts incorrect, missing, and extra words against the reference transcript, so lower is better.

  • First text arrives after 1.70s at the median, and the transcript is final 162ms after the speaker stops. Both are measured on the same 206 streamed turns as every other model, and the page shows P90 and P95 as well, which is what a caller meets on a bad turn.

  • Every model runs the same two datasets under the same conditions: 1,000 public clips from Pipecat's STT benchmark dataset for accuracy at scale, and eight licensed conversation recordings from Ocular for real speech. Latency comes from 206 streamed turns, using each provider's own supported finalization contract. Zoom Scribe Pro returned usable text for 1,000 of the public clips.