You're looking at a screenshot of a menu in Japanese, or a sign in Russian, or a PDF in Arabic. You need the text — not a photo, not a translation app's camera view, but actual selectable text you can paste into a document or translator. On a Mac, the built-in OCR tools handle some languages beautifully and others not at all. Here's what actually works.

Apple Vision, the framework behind Live Text and apps like Cheese! OCR, recognizes a specific set of languages. As of macOS 14, that set includes English, Simplified Chinese, Traditional Chinese, Japanese, Korean, French, German, Italian, Spanish, and Portuguese. macOS 12 started with English, French, Italian, German, Portuguese, Spanish, and Chinese; Japanese and Korean arrived in macOS 13. The system auto-detects the language — you never select one manually. Mixed-language text, like English embedded in a Japanese page, is handled automatically.

What Apple Vision can't read

Apple Vision does not recognize Cyrillic scripts (Russian, Ukrainian), Arabic, Hebrew, Thai, or Vietnamese as of early 2026. Right-to-left scripts like Arabic and Hebrew require RTL-aware OCR, which Apple Vision doesn't support. If your screenshot contains any of these, Live Text and Cheese! OCR will return nothing useful.

For unsupported scripts, you have two practical options. Tesseract is an open-source CLI tool you can install via Homebrew. You'll need to download the appropriate language data files — tesseract-lang covers most scripts. It's free and runs locally, but the accuracy varies and setup takes a few minutes. Cloud OCR services like Google Cloud Vision or Azure AI Vision support a much wider range of languages and generally produce cleaner results, but they send your image to a remote server.

After OCR: translation paths

Once you have the text, you need to translate it. Apple Translate is built into macOS 12+ and does on-device translation for many language pairs without sending anything to the cloud. It's fast and private, but the language coverage is narrower than cloud alternatives. DeepL is generally the best choice for European languages — it handles nuance better than most competitors. Google Translate has the broadest coverage if you need something more obscure like Thai or Vietnamese.

For Chinese, Apple Vision auto-handles both Simplified and Traditional script — no toggle needed. For Japanese, it handles hiragana, katakana, and printed kanji reliably; freehand handwriting recognition is partial.

When the built-in tool is enough

If you only OCR occasionally and only in the languages Apple Vision supports, Live Text is free, fast, and right there. Stop reading and use the tool you already have. The same applies if you're working with mixed-language text that stays within the supported set — Apple Vision handles it automatically.

If you need Cyrillic, Arabic, Hebrew, Thai, or Vietnamese, or if you need to batch-process many screenshots, you'll need something else. Cheese! OCR is built on Apple Vision and inherits the same language list — it won't help with unsupported scripts. For those cases, Tesseract or a cloud service is your actual path forward.

A short checklist

  1. Check if your target language is in Apple Vision's supported list (English, Chinese, Japanese, Korean, French, German, Italian, Spanish, Portuguese).
  2. If it is, use Live Text or Cheese! OCR — no setup needed.
  3. If it isn't, install Tesseract via Homebrew and download the language data for your script.
  4. For higher accuracy on unsupported scripts, use Google Cloud Vision or Azure AI Vision.
  5. After OCR, translate with Apple Translate (on-device), DeepL (best for European languages), or Google Translate (broadest coverage).
  6. If you batch-process screenshots regularly, consider a dedicated tool like Cheese! OCR for the supported languages, and a separate workflow for the rest.