Translate English to Chinese Озод
Fast, accurate English to Chinese translation powered by AI. No sign up required.
Дар бораи ин тарҷума
Our English to Chinese translator uses the MadLAD-400 AI model, which supports over 450 languages. It provides fast, accurate translations for documents, messages, and any text you need converted.
- Бо истифода аз Google's MadLAD-400 3B модел (Apache 2.0)
- Ҳамчунон хуб ҳуҷҷатҳои дароз ва ибораҳои кӯтоҳро идора мекунад
- Бе маҳдудияти аломатҳо дар дараҷаи озод
- Махфӣ ва бехатар -- матни шумо захира намешавад
How to Translate English to Chinese
- Go to the Free.ai Translator
- Select English as the source language
- Select Chinese as the target language
- Матнро чоп кунед ё часпонед
- & Тарҷума
Тафсилоти тарҷума
| Аз | English |
| To | Chinese |
| Коди манбаъ | en |
| Рамзи ҳадаф | zh |
| Намунаи AI | МадЛАД- 400 3Б |
| & Нарх | Озод |
Тарҷумаҳои иловагӣ
Намоиши ҳамаи ҷуфтҳои забонСаволҳои зиёд
English → Chinese translation on Free.ai runs through NLLB-200 (Meta's 200-language multilingual model) for the default route, with Madlad-400 as a fallback on long-tail vocabulary. English and Chinese are typologically distant (different writing systems, different word orders, very different grammar), so quality on literary or idiomatic content lags closer language pairs — factual prose and technical text still come through cleanly. The form on this page lets you paste up to ~10,000 characters; for longer text upload a file or use the /v1/translate/ API. If a passage looks off, regenerate — the engine samples differently each run and the second pass often catches a missed nuance.
Yes — English → Chinese translation is free for the bulk of normal use. Anonymous users get a per-IP daily pool; signed-in free accounts get a larger daily pool (~50,000 characters/day) plus 10,000 signup tokens. Beyond that, top-ups start at $1 (750,000 tokens, ~150,000 characters of English) — no subscription, no monthly minimum.
Numbers stay numeric, but locale conventions shift — decimal separators, thousand separators, and date order are localized to Chinese (zh) where it is unambiguous. Currency symbols are preserved as written (the engine does not convert amounts). For i18n catalogs, upload a JSON file and the translator will leave ICU placeholders ({count}, {name}) untouched.
Yes. /translate/ accepts DOCX, PDF, TXT, MD, SRT (subtitle files), and JSON (i18n catalogs). Drop the file on the upload zone and pick Chinese (zh) as the target — the translated file downloads with the same layout, fonts, and (for DOCX/PDF) inline images preserved in place. JSON keys are kept; only values are translated. SRT timestamps are preserved exactly.
English → Chinese runs on NLLB-200 (Meta's 200-language multilingual model) for the default route, with Madlad-400 as a fallback on long-tail vocabulary. These models are self-hosted on Free.ai GPUs — your text never leaves our infrastructure for the default route. Premium routes (GPT-4o, Claude, Gemini) are available for pairs and contexts where the frontier model materially beats the open-source engines; those route through the upstream provider under our DPA. The route used for any given request is shown in the API response under "model".
Yes. Chinese (zh) is rendered in its native script with full-width CJK punctuation (、。「」) where idiomatic, and the engine preserves CJK-specific line-break rules in DOCX exports. Copy-paste into any modern editor works directly; older terminals may need a CJK-aware font installed.
Translation cost on Free.ai is ~5 tokens per character of input. $1 buys 750,000 tokens, which covers roughly 150,000 characters of English input (about 25,000 words, or a 50-page document). A $5 top-up covers ~750,000 characters. Free accounts get a daily pool sized at ~50,000 characters/day plus the 10,000-token signup bonus, so short and medium documents are typically free. The cost is computed on the English input length — the Chinese output length does not affect billing.
Yes — drop an .srt file on /translate/ and pick Chinese (zh) as the target. The translator preserves every cue number and timestamp exactly; only the dialogue text is rewritten. For long films, the file streams through chunk-by-chunk so a 2-hour movie subtitle file still finishes in well under a minute. If you need to dub the translated track as audio, pipe the output into /voice/tts/ — pick a Chinese voice and you get a synced narration. For the reverse flow (audio file → translated subtitles), use /transcribe/ first, then this translator.
Idioms are the honest weak spot. Common ones ("piece of cake", "break a leg") usually map to the Chinese equivalent because they appear in the training corpus; novel or regional slang often gets literal-translated. If accuracy on idioms matters (marketing copy, fiction dialogue), drop a footnote in your prompt — "translate idiomatically, prefer natural Chinese" — and the model will look harder for an equivalent expression instead of a word-for-word rendering. For legal, medical, or code-heavy text, the engine is conservative on purpose — proper nouns and technical terms are left as-is or rendered in their conventional form.
Yes. POST to /v1/translate/ with source="en", target="zh", and text="..." in the JSON body, plus your Bearer token. The response returns {"translation": "...", "detected_source": "...", "model": "..."}. For long documents, send file_url= instead of text= and the API will fetch, chunk, and stitch the result. Full reference and SDK snippets at /api/.
Yes — the reverse direction lives at /translate/zh-to-en/ and uses the same engine routing in the opposite direction. Quality can differ asymmetrically: one direction may read more naturally than the other because the training corpus is rarely balanced. If you need a round-trip sanity check (translate English → Chinese → English again), do it manually in two passes — automatic round-trip is intentionally not exposed because it tends to mask errors rather than catch them.
Honest answer: on common English → Chinese prose, Free.ai's NLLB-200 route lands within a few BLEU points of Google Translate and DeepL on most public benchmarks. Google still has the edge on rare named entities (mid-tier brands, niche place names) because of its larger web index; DeepL often wins on long European-language documents thanks to its document-level conditioning. Free.ai's advantages: no character cap on uploaded files, no daily request quota for signed-in users, a real API that doesn't throttle, and SRT / DOCX / PDF support without an enterprise plan. Try the same paragraph in both — /translate/ never charges you to compare.