OCR

OCR pulls text out of an image or PDF. Items OCR-01 to OCR-05 describe a printed page, ruq'ah handwriting, a poor scan, an invoice, and a vocalised text. Sample files are not in the repo yet, so there is no character-error rate. If a source shows an Arabic example, such as Mistral's launch post, we quote it as a vendor claim, not a result. Prices are often usage-based and may be unverified.

Model All FreeFreemiumPaidOpen sourceEnterpriseNot set in the table Arabic-claim status All The source mentions Arabic supportNo Arabic claim on the pageThe page could not be verifiedLocalized docs or site UI onlyA number the vendor published about itself A free option is mentioned All Mentioned Not mentioned Sort Name First wave first Score (unavailable) Clear filters

shown

Questions

Can they turn an image into a Word file that matches the original?

That is what the test will measure once samples exist. There is no accuracy figure today.

Thursday letter

Company website

Email Letter language العربية English I agree to receive the weekly AI Arabi Hub letter by email, and I can unsubscribe at any time. Privacy policy Subscribe