On-device OCR, even offline
Apple Vision and CoreML recognize text right on your device — no upload, no account. Tesseract is built in as a fallback, and Gemini cloud OCR is there only if you choose it.
OCREditor
OCREditor recognizes text on-device with Apple Vision, lets you rewrite it in place, and rebuilds the background — all privately on your iPhone, iPad or Mac.
Apple Vision and CoreML recognize text right on your device — no upload, no account. Tesseract is built in as a fallback, and Gemini cloud OCR is there only if you choose it.
Tap any text block to edit, delete or add a new region. Set separate CJK and Latin fonts so mixed-language text stays consistent.
Erased text is filled with a fast 4-corner gradient patch, or with the optional AI LaMa model for complex backgrounds — cached locally after the first download.
Edits and exports are generated on your device. No ad SDKs, no tracking, and nothing is uploaded unless you explicitly enable cloud OCR.
On iOS/iPadOS, ensure you are using the packaged app so Apple Vision is available. On macOS desktop browser, verify ocr_server.py is running on 127.0.0.1:5001, or switch to browser Tesseract OCR.
The AI LaMa model downloads once and caches locally. If WebGPU/WASM memory is limited on your device, OCREditor automatically falls back to the fast 4-corner bilinear background patch pipeline.
Use Preset Fonts to set separate Primary (CJK) and Secondary (Latin/Numbers) fonts, and enable Force Computer Font if you want to override original image fonts consistently.
No. OCREditor works 100% offline with Native OCR and Tesseract. A Gemini API key is only needed if you choose to use optional Gemini Cloud OCR.
All exports are generated locally and saved to the folder you pick (or your browser/device Downloads folder).
Edit the text inside any image.