Universal 3 Pro
AssemblyAI
Speech to textTranscribes a recording into words
- Type
- Proprietary
- Input
- None held
- Output
- None held
- Cached
- None held
We don't hold a list price for this model yet · hosted only — no weights published
Our take
Written Aug 2, 2026Universal 3 Pro is AssemblyAI's hosted-only speech-to-text model that converts audio to written words across 99 languages. It excels on clean, structured English audio with roughly one word in ninety wrong, but accuracy drops sharply on meetings, podcasts and accented speech.
Choose this for high-stakes financial transcription where measured accuracy is strongest, or clean read-aloud applications with minimal background noise. Consider it when you need broad language coverage, though accuracy beyond English is unverified in our data. Skip it if you are transcribing meetings, podcasts or heavily accented speech, where error rates rise tenfold; or if you need transparent pricing and availability, which are undisclosed.
The case for it
- Exceptional accuracy on clean, structured audio: roughly one word in ninety wrong on read speech, and similar performance on financial calls.
- Broad language coverage with 99 languages supported.
The case against it
- Severe accuracy collapse on conversational and accented audio: error rates rise roughly tenfold on recorded meetings and accented speech compared with clean read speech.
- No pricing or availability data: zero offers in catalogue and no costs disclosed.
- Every accuracy figure is English only; performance across the other 98 languages is unverified in our data.
How good is it?
TranscriptionTurning speech into text4 of 5Open ASR WER · 21st of 74
94.8%
Misses roughly one word in 19, averaged over nine English test sets.
99
Stated by the leaderboard; we do not hold the list itself.
Percentage of words wrong on each set, lower better. Bars are scaled to this model's own worst case, not to the board.
The figures above come from the Open ASR Leaderboard, an independent public test that runs every model on the same recordings. It is the only measurement of transcription quality we know of, so there are no other scores to show.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done, which is why they get no rating.
Every published score for this model8 scoresEvery figure we hold, from 8 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Where to get it
We hold no priced listing for Universal 3 Pro.
There is no copy to download and no host in our price data, so AssemblyAI is where to look. We watch OpenRouter, the provider APIs we track and the LiteLLM price set; this version appears in none of them, which is a gap in what we collect rather than a statement about what AssemblyAI sells.
Models people weigh against Universal 3 Pro
When we formed this view
Dates behind this page
Prices last checked 6h ago
What we do not know about this model yet
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- We don't hold a list price for this model yet — the gap is ours, not the lab's.
Licence and identifiers
What the licence allowsWe hold no licence record for this model. Inside are the identifiers you need to pull it — its Hugging Face repo where we have one, our slug and a machine-readable card.
Licence
Commercial API terms. We hold no licence record for this model, so there is nothing to summarise here.
Identifiers
- Modality record
- audio->text
- Catalogue slug
- assemblyai-universal-3-pro