Runemic OCR Arena

Every OCR engine. One page.

Drop an image or a PDF, pick up to three OCR engines and compare what each one reads, side by side. They run right here in your browser, on your own CPU or GPU: your pages never leave your device.

1

Your page

No page handy? Try one:
2

Engines pick up to 3

Your CPU runs in your browser Your GPU · WebGPU runs in your browser, on your graphics card GPU server · soon large models, coming soon API sent to a server to read Instant no OCR needed
Coming soon: large OCR models on a GPU server
Choose a page and at least one engine.

What runs where

There is no server behind this Space: the engines download once into your browser (and stay cached) and read your pages on your own device. Only the Runemic engine sends a page out, to read it with Runemic's models.

EngineByRuns onDownloadLicence

Private by design

PaddleOCR, EasyOCR, Tesseract, GLM-OCR, LightOnOCR and the PDF text layer read your page inside this browser tab. Nothing is uploaded and nothing is stored. The Runemic engine is the one exception: it sends the page to api.runemic.com, which reads it and doesn't keep it.

Reading thousands of pages, or Persian, Urdu and old scans that need to be right? Runemic's API reads them in one call, with a free credit to start.