Qwen3-ASR 0.6B
Qwen3-ASR 0.6B is a 0.6B-parameter speech-to-text model from Qwen, released under the Apache-2.0 license. The reviewed source lists 30 languages + 22 Chinese dialects (30 explicitly enumerated in our record), and the model is designed for batch transcription, with live use only through a runtime or chunking wrapper. Among the common setups we test, it ranks highest for Japanese live transcription (#4 of 26 in the matcher).
Modern multilingual ASR with official vLLM streaming inference. The current streaming implementation re-feeds accumulated audio rather than using a cache-based streaming encoder, and it does not return timestamps. Detailed timestamps use the separate Qwen3 Forced Aligner.
Key facts
| Provider | Qwen |
|---|---|
| Model family | Qwen3-ASR |
| Parameters | 0.6B |
| License | Apache-2.0 · Permissive — license text ↗ |
| Language coverage | 30 languages + 22 Chinese dialects · 30 enumerated in our record |
| Streaming | Streaming via a runtime or chunking wrapper |
| Timestamps | Via a separate alignment component |
| Speaker diarization | No — needs a separate component |
| Speech translation | No |
| Documented runtimes | qwen-asr package, vLLM, Docker image qwenllm/qwen3-asr — how to run it |
| Primary source | github.com/QwenLM/Qwen3-ASR ↗ |
| Last reviewed |
Hardware and workload fit
These are Speech Model Match's conservative 1–5 editorial ratings, relative to the other 19 reviewed models. They come from model size, official memory guidance and documented runtimes — not from a universal benchmark. How the ratings work.
| Dimension | Rating | Scale |
|---|---|---|
| CPU-only fit | 2/5 | |
| Apple Silicon fit | 3/5 | |
| NVIDIA GPU fit | 5/5 | |
| Speed | 4/5 | |
| Accuracy (relative) | 5/5 | |
| Low memory use | 3/5 | |
| Long recordings | 4/5 | |
| Live / low latency | 4/5 | |
| Subtitle work | 4/5 | |
| Edge / low-power devices | 2/5 |
Where it is strong
- NVIDIA GPU fit: rated 5/5
- Speed: rated 4/5
- Accuracy (relative): rated 5/5
- Long recordings: rated 4/5
- Live / low latency: rated 4/5
- Subtitle work: rated 4/5
What to check first
- CPU-only fit: rated 2/5
- Edge / low-power devices: rated 2/5
- No native speaker diarization — pair it with a separate diarization pipeline if you need speaker labels
How it ranks in common setups
Ranks come from the same deterministic matcher that runs on the homepage, applied to all 26 reviewed models. They are decision guidance, not accuracy claims.
| Setup | Rank | Fit score | Top match |
|---|---|---|---|
| Long recordings on an 8–12 GB NVIDIA GPU, balanced | #8 of 26 | 75/100 · Strong fit | Whisper turbo |
| Live captions or voice input (native streaming + timestamps, speed first) | #10 of 26 | 79/100 · Strong fit | Nemotron 3.5 ASR Streaming 0.6B |
| Subtitles with timing, accuracy first | #7 of 26 | 80/100 · Strong fit | Whisper medium |
| Meetings with speaker labels (diarization + timestamps) | #8 of 26 | 77/100 · Strong fit | VibeVoice-ASR 7B |
| CPU-only machine, low memory use | #17 of 26 | 58/100 · Possible fit | VibeVoice-ASR-BitNet |
| Apple Silicon with 8–16 GB unified memory, long recordings | #16 of 26 | 66/100 · Possible fit | Whisper small |
| Japanese live transcription (native streaming, speed first) | #4 of 26 | 77/100 · Strong fit | Nemotron 3.5 ASR Streaming 0.6B |
Supported languages
The official model card also lists 22 Chinese dialects in addition to the languages below.
- Arabic
ar - Cantonese
yue - Czech
cs - Danish
da - Dutch
nl - English
en - Filipino
fil - Finnish
fi - French
fr - German
de - Greek
el - Hindi
hi - Hungarian
hu - Indonesian
id - Italian
it - Japanese
ja - Korean
ko - Macedonian
mk - Malay
ms - Mandarin Chinese
zh - Persian
fa - Polish
pl - Portuguese
pt - Romanian
ro - Russian
ru - Spanish
es - Swedish
sv - Thai
th - Turkish
tr - Vietnamese
vi
Compare coverage across every reviewed model in the language support matrix.
Community activity on Hugging Face
- Downloads, last 30 days
- 584,234
- Likes
- 399
- Repository last updated
- 30 Jan 2026
- Published on the Hub
- 28 Jan 2026
Figures from the Hugging Face Hub API, 11 Nov 2026. Source: huggingface.co/Qwen/Qwen3-ASR-0.6B ↗. Downloads count file downloads, not active users, and do not affect the matcher's scores. Compare all models.
How to run Qwen3-ASR 0.6B locally
Qwen publishes a qwen-asr Python package that loads every released Qwen3-ASR model, with vLLM for high-throughput serving. Runtimes below are the ones the official model card or repository documents for this model (checked November 2, 2026).
- qwen-asr package ↗Official package · Transformers backend
- vLLM ↗Serving engine · required for streaming
- Docker image qwenllm/qwen3-asr ↗Official package · needs the NVIDIA Container Toolkit
Install command shown in the official github.com page ↗:
pip install -U qwen-asr
Audio input: Audio can be a local path, URL, base64 string or NumPy array with sample rate. Streaming output is only available with the vLLM backend.
The official README and model card describe NVIDIA GPU deployment only; no CPU or Apple Silicon path is documented.
Whatever runtime you choose, measure speed and accuracy on your own audio with the benchmark tool; results differ a lot between runtimes, precision and hardware.
Before you deploy Qwen3-ASR 0.6B
A printable checklist built from the facts on this page. Tick it off before putting the model into production.
- Language. Confirm your language is in the 30-language list on this page, and test any accents or code-switching you expect.
- License. Apache-2.0: Give recipients a copy of the license, keep copyright and attribution notices, carry forward any NOTICE file and mark files you changed. The patent license ends if you start patent litigation over the work. License guide.
- Install. Start from the official command:
pip install -U qwen-asr(full run notes). - Audio input. Audio can be a local path, URL, base64 string or NumPy array with sample rate. Streaming output is only available with the vLLM backend.
- Hardware. No vendor memory figure is published; our ratings are CPU 2/5, Apple 3/5, NVIDIA 5/5.
- Features you rely on. For Qwen3-ASR 0.6B, streaming only through a runtime or chunking wrapper; timestamps need a separate aligner; speaker labels need a separate diarization component. Test each one you need with real audio.
- Measure. Run 10–20 representative clips, score them with the WER calculator and record speed in the benchmark workspace before you commit.
Alternatives and related models
Same family: Qwen3-ASR 1.7B.
Similar fit profile: Nemotron 3.5 ASR Streaming 0.6B (NVIDIA, 600M), Parakeet TDT 0.6B v3 (NVIDIA, 600M), Distil-Whisper large-v3 (Hugging Face / Distil-Whisper, large-v3 distilled). Similarity is computed from the ten fit ratings plus streaming and diarization support.
| Fact | Qwen3-ASR 0.6B | Qwen3-ASR 1.7B |
|---|---|---|
| Provider · size | Qwen · 0.6B | Qwen · 1.7B |
| License | Apache-2.0 Permissive | Apache-2.0 Permissive |
| Languages | 30 languages + 22 Chinese dialects | 30 languages + 22 Chinese dialects |
| Streaming | Via runtime | Via runtime |
| Timestamps | Separate aligner | Separate aligner |
| Speaker diarization | Separate pipeline | Separate pipeline |
| CPU · Apple · NVIDIA fit | 2/5 · 3/5 · 5/5 | 1/5 · 3/5 · 5/5 |
| Live · long recordings | 4/5 · 4/5 | 3/5 · 5/5 |
| Main runtime | qwen-asr package | qwen-asr package |
Head-to-head pages: Qwen3-ASR 0.6B vs Fun-ASR-MLT-Nano-2512
Compare Qwen3-ASR 0.6B side by side
Guides that cover Qwen3-ASR 0.6B
- Qwen3-ASR vs Whisper: which local ASR family fits better?
- Best local STT models for an 8 GB GPU
- Best local speech-to-text models for streaming
- Best local speech-to-text models for multilingual use
- Best local speech-to-text models in 2026
- Best local Whisper alternatives in 2026
Check Qwen3-ASR 0.6B for your setup
Pick your language, hardware and workload to see this model's fit score, the reasons behind it and where it ranks against all reviewed models. Everything runs in your browser.
Common questions
Can I use Qwen3-ASR 0.6B commercially?
Permissive license; commercial use is allowed under the listed terms. The listed license is Apache-2.0; read the current license text ↗ before shipping a product, because terms can change. The license guide compares the obligations of all 26 reviewed models.
Does Qwen3-ASR 0.6B support real-time streaming?
Not natively. Live use relies on a runtime or chunking wrapper around a batch model, which adds latency and can split words at chunk borders. If live latency is critical, compare it with a native streaming model such as Nemotron 3.5 ASR Streaming 0.6B.
Which languages does Qwen3-ASR 0.6B transcribe?
30 languages + 22 Chinese dialects. The full list of 30 explicitly enumerated languages is shown above.
Can Qwen3-ASR 0.6B tell speakers apart?
Not natively. Combine it with a dedicated diarization component and evaluate that component separately; the meetings guide covers this workflow.
Primary sources
Reviewed against primary sources on September 16, 2026. Model behavior, runtime support, language coverage and licenses can change; re-check the linked primary source before production use. See the full source directory.