Qwen3 ASR 0.6B
Qwen · released Jan 28, 2026 · Qwen/Qwen3-ASR-0.6B
Speech to textTranscribes a recording into words
- Type
- Open weightsApache License 2.0
- Languages
- 52
- Size
- 0.9B
Context measured in tokens
Our take
Written Sep 4, 2026Qwen3 ASR is a tiny downloadable speech-to-text model with an Apache licence and 52 languages. It is unusually accurate on messy real-world audio for its size, though it lags on clean read-aloud speech where larger models excel.
Pick this for podcasts, video or everyday internet audio where it beats most rivals, or for meeting transcription on limited hardware at 439 times real time. Use it for multilingual deployments needing a permissive licence, or accent-heavy recordings where it punches above its weight. Skip it if your audio is clean studio speech, or if you need a hosted provider rather than self-hosting.
The case for it
- Unusually accurate on messy real-world audio: podcasts, meetings and accented speech all better than most of 74 models.
- Extremely fast: an hour of audio in about eight seconds on standard leaderboard hardware.
- 52 languages under an Apache 2.0 licence, broad coverage for a downloadable transcription model.
- Strong on accented and non-native speech, including European parliamentary recordings and multi-country earnings calls.
The case against it
- Worse than most on clean read-aloud audio, the easiest case, at 1.69% word error rate against a 0.9% field best.
- No commercial hosting options available; you must run it yourself.
- Small parameter count may limit fine-tuning headroom.
How good is it?
An open transcription model for turning podcasts and video audio into written text.
- transcribing podcasts and video audioPodcasts and video · 18th of 92
TranscriptionTurning speech into text
439×51st of 74
an hour of audio in 8 seconds, on the board's own hardware. Your machine will differ.
52
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; 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 model7 scoresEvery figure we hold, from 7 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. 21 GB spare means a 10% error in the size would not change the answer.
Apple M1 Pro (16-core GPU) · 32 GB
Room to spare. 22.2 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. 4.2 GB spare means a 10% error in the size would not change the answer.
These cards answer whether Qwen3 ASR 0.6B 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.
Models people weigh against Qwen3 ASR 0.6B
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 allowsApache License 2.0, 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
Apache License 2.0
Fully permissive: commercial use, redistribution, and derivatives allowed. Requires attribution and a copy of the license. Includes an express patent grant.
Identifiers
- Hugging Face
- Qwen/Qwen3-ASR-0.6B
- Architecture
- Dense
- Takes in, gives back
- Audio in, text out
- Catalogue slug
- qwen-qwen3-asr-0-6b