Canary Qwen 2.5b
NVIDIA · released Jun 26, 2025 · nvidia/canary-qwen-2.5b
Speech to textTranscribes a recording into words
- Type
- Open weightsCreative Commons Attribution 4.0
- Languages
- 1
- Size
- 2.6B
Context measured in tokens
Our take
Written Sep 11, 2026Canary Qwen 2.5b is a compact downloadable speech-to-text model from NVIDIA that turns English audio into written words at 867 times real-time speed. It performs well on clean and professional recordings, though its accuracy drops on harder audio and only English is measured in our data.
Pick this for fast batch transcription of English audio where speed matters — it processes an hour of audio in about four seconds. Use it for clean read-aloud or financial calls, where it gets roughly one word in eighty wrong, or for European-accented English at about one in twenty-five. Skip it if you need languages beyond English, speaker separation, timestamps, or hosted inference without setting up your own hardware.
The case for it
- Among the faster recognisers we list at 867× real time — an hour of audio in roughly four seconds on reference hardware.
- Strong on clean and professional English: about 1.2% word error on read-aloud and 1.7% on financial calls, better than most models.
- Respectable on harder everyday audio: 7.8% on podcasts and video, 7.9% on meetings, both better than most.
- Creative Commons Attribution 4.0 licence allows commercial use with attribution.
The case against it
- Only English is measured in our data; no non-English accuracy figures are held.
- Compact size shows on the hardest audio: 6.4% on accented earnings calls and 7.9% on meetings, several times its clean-audio rate.
- No hosted inference options currently listed; you must run it yourself.
How good is it?
An open speech-to-text model for turning recordings of meetings, accented speakers and general speech into written text.
- turning spoken English into written textOpen ASR WER · 16th of 76
- transcribing recordings of meetings in a roomRecorded meetings · 21st of 92
- transcribing speakers with a range of accentsAccented speech · 19th of 76
TranscriptionTurning speech into text4 of 5Open ASR WER · 16th of 76
95.6%
Misses roughly one word in 23, averaged over nine English test sets.
867×34th of 74
an hour of audio in 4 seconds, on the board's own hardware. Your machine will differ.
1
Listed on the model card. The accuracy above is English only.
Percentage of words wrong on each set, lower better. Bars are scaled to this model's own worst case; the placing beneath each rate is against every model measured on that set.
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.
Each of these is the same transcription job on a different kind of recording, so together they say where it holds up and where it slips — not how closely it follows an instruction.
Every published score for this model9 scoresEvery figure we hold, from 9 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
Comfortable fit
GeForce RTX 4090 · 24 GB
Room to spare. 19.9 GB spare means a 10% error in the size would not change the answer.
GeForce RTX 5090 · 32 GB
Room to spare. 27.9 GB spare means a 10% error in the size would not change the answer.
Apple M2 (8-core GPU, 8GB unified) · 8 GB
Room to spare. 3.1 GB spare means a 10% error in the size would not change the answer.
These cards answer whether Canary Qwen 2.5b loads, not how fast it transcribes. Throughput figures for a transcription model come from its text decoder, so treat this as a fit answer rather than a speed one.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first. This is a fit answer: whether it loads, not how fast it transcribes.
When we formed this view
Recent changes
What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
Licence and identifiers
What the licence allowsCreative Commons Attribution 4.0, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Creative Commons Attribution 4.0
Permissive content license: any use with attribution. Common for datasets and some model weights.
Identifiers
- Hugging Face
- nvidia/canary-qwen-2.5b
- Takes in, gives back
- Audio in, text out
- Catalogue slug
- nvidia-canary-qwen-2-5b