GPT-4o Transcribe
GPT-4o Transcribe is 12th of 15 on the Pipecat Dataset and 15th of 15 on final-text delay.
- 3.90%
- Pipecat WER · 12th
- 12.45%
- Ocular WER · 15th
- 652ms
- Final text · 15th
Quality and speed frontier
Every model's error rate on the public clips, against how long its transcript takes to settle.
GPT-4o Transcribe against every other model
Open on the metric GPT-4o Transcribe places highest on, with the other three a tab away. Lower is better on all of them, so the order runs best first.
Word error rate on 1,000 clips from Pipecat's STT benchmark dataset. Lower is better.
| # | Model | Pipecat Dataset | clips | |
|---|---|---|---|---|
| 1 | AssemblyAI Universal 3.5 Pro | 1.93% | 1,000 | |
| 2 | Google Chirp 3 | 2.00% | 1,000 | |
| 3 | Reson8 | 2.10% | 1,000 | |
| 4 | GPT Realtime Whisper | 2.16% | 1,000 | |
| 5 | Cartesia Ink 2 | 2.39% | 1,000 | |
| 6 | Speechmatics Linden | 2.46% | 998 | |
| 7 | Google Chirp 2 | 2.59% | 1,000 | |
| 8 | Inworld STT-1 | 2.72% | 1,000 | |
| 9 | GPT-4o Mini Transcribe | 3.35% | 1,000 | |
| 10 | Deepgram Nova-3 | 3.46% | 1,000 | |
| 11 | Smallest Pulse | 3.66% | 1,000 | |
| 12 | GPT-4o Transcribe | 3.90% | 1,000 | |
| 13 | Deepgram Flux English | 3.90% | 999 | |
| 14 | Deepgram Flux Multilingual | 4.96% | 999 | |
| 15 | Gradium | 6.48% | 1,000 |
Counts are the clips each model returned usable text for, so they differ between models and between datasets.
First text and final text, at three percentiles
Measured on the same 206 streamed Ocular Dataset turns as every other model. P90 and P95 say what a caller meets on a bad turn, which a median hides.
| Measure | P50 | P90 | P95 |
|---|---|---|---|
| Time to first text | No text until the speaker stops | ||
| Final-text delay | 652ms | 1.73s | 1.97s |
GPT-4o Transcribe on the other two views
The same charts the benchmark page carries, with GPT-4o Transcribe ringed. Consistency across the two datasets is one reading; how fast a model starts against how fast it commits is the other.
Pipecat vs Ocular accuracy
Every model's error rate on the public clips, against the same model on real conversations.
Transcription response time
How soon each model returns its first words, against how soon its transcript is final.
What the numbers rest on
- 1,000
- Pipecat clips scored
- 8
- Ocular recordings scored
- 206
- turns timed
- 0
- requests that failed
A clip counts only where the model returned usable text, so counts differ between models. Word error rate is measured the same way throughout: incorrect, missing, and extra words against the reference transcript.
Frequently asked questions
Where does GPT-4o Transcribe place in the benchmark?
GPT-4o Transcribe places highest on the Pipecat Dataset, 12th of 15 at 3.90%. Its weakest placing is final-text delay, 15th of 15 at 652ms.
How accurate is GPT-4o Transcribe?
GPT-4o Transcribe transcribes the 1,000 public Pipecat clips at 3.90% word error rate and the eight licensed Ocular recordings at 12.45%. Word error rate counts incorrect, missing, and extra words against the reference transcript, so lower is better.
Why is there no time to first text for GPT-4o Transcribe?
GPT-4o Transcribe returns no text at all while the speaker is talking, so a measured time to first text would track how long each turn happened to be rather than anything about the model. Its final-text delay is still comparable and is reported in full.
How was GPT-4o Transcribe measured?
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. GPT-4o Transcribe returned usable text for 1,000 of the public clips.