Chinese translation and speech-to-text models zh
1 model into Chinese, 12 from Chinese, 14 speech-to-text. Best scored into Chinese: English to Chinese (chrF++ 23.1). Best scored from Chinese: Chinese to English (chrF++ 51.5).
Which open models work for Chinese?
Windy Translate lists 1 open translation model into Chinese and 12 from Chinese. All of them can be downloaded from Hugging Face and run on your own hardware. By licence: 11 Apache-2.0 and 2 CC-BY-4.0 (terms and attribution text: licensing). 12 of them have a screening score on FLORES-200. The best scored into Chinese is translate-en-zh (chrF++ 23.1, Limited). The best scored from Chinese is translate-zh-en (chrF++ 51.5, Good).
With English. English to Chinese: translate-en-zh (chrF++ 23.1, Limited). Chinese to English: translate-zh-en (chrF++ 51.5, Good). To translate between Chinese and a language with no direct model, you can go through English with these two.
By band, the scored Chinese models are 1 Good, 4 Usable, 5 Limited and 2 Not recommended (Excellent is chrF++ 60 or more, Good 45 to 60, Usable 30 to 45, Limited 15 to 30). A score is a 48-sentence screening of news and encyclopedia text, not a measure of accuracy, and a difference of a point or two is noise. How we measured.
Speech-to-text. 2 speech-to-text models made for Chinese: listen-windy-lingua-zh and listen-windy-lingua-zh-ct2. Also, Whisper's tokenizer lists Chinese as one of its languages, so all 12 multilingual Whisper-based models here can transcribe it: see the speech models. Accuracy differs a lot between languages and model sizes, and this site has not measured speech accuracy.
Each model page has a short snippet to download and run the model yourself, its licence and attribution text, and its score where there is one. Pick the best scored model for your direction, then try it on a few sentences of your own text: a screening score is a starting point, not a guarantee.
Into Chinese
| Model | chrF++ (FLORES-200 dev) | Licence |
|---|---|---|
| English to Chinese best scored translate-en-zh | 23.1 Limited | Apache-2.0 |
From Chinese
| Model | chrF++ (FLORES-200 dev) | Licence |
|---|---|---|
| Chinese to English best scored translate-zh-en | 51.5 Good | CC-BY-4.0 |
| Chinese to Italian translate-zh-it | 34.8 Usable | Apache-2.0 |
| Chinese to Bulgarian translate-zh-bg | 34.4 Usable | Apache-2.0 |
| Chinese to Dutch translate-zh-nl | 30.7 Usable | Apache-2.0 |
| Chinese to Hebrew translate-zh-he | 30.2 Usable | Apache-2.0 |
| Chinese to Finnish translate-zh-fi | 29.3 Limited | Apache-2.0 |
| Chinese to German translate-zh-de | 28.3 Limited | Apache-2.0 |
| Chinese to Vietnamese translate-zh-vi | 28.1 Limited | Apache-2.0 |
| Chinese to Swedish translate-zh-sv | 27.4 Limited | Apache-2.0 |
| Chinese to Ukrainian translate-zh-uk | 10.9 Not recommended | Apache-2.0 |
| Chinese to Japanese translate-tc-big-zh-ja | 1.4 Not recommended | CC-BY-4.0 |
| Chinese to Malay translate-zh-ms | not in FLORES-200 | Apache-2.0 |
Chinese speech-to-text
| Model | Formats | Licence |
|---|---|---|
| listen-windy-lingua-zh | safetensors | Apache-2.0 |
| listen-windy-lingua-zh-ct2 | ct2-int8 | Apache-2.0 |
Scores: chrF++ on FLORES-200 dev, 48 sentences, measured 2026-08-04. Method.