Arabic Image to Text
Free Arabic OCR — right-to-left, connected script, measured honestly.
VisionParse converts Arabic images to editable UTF-8 text free in your browser, reading the connected right-to-left script with a dedicated Arabic model.
Our measured figure is 93.84% character accuracy on a printed paragraph that includes Arabic-Indic numerals (٢٤١٧, ٢٠٢٦) — the digits invoices actually carry, and a known weak point for OCR engines generally. We publish the real number rather than rounding it up: most errors land on dots and numeral shapes, not on word structure.
- Free · no account
- JPG · PNG · WEBP · TIFF · BMP · HEIC
- 3 MB · 3 pages
- Deleted within 1h
Drop an image to read it
JPG, PNG, WEBP, TIFF, BMP, HEIC — up to 3 MB. Your file is read on our servers in Frankfurt and deleted within the hour.
What this tool does, in numbers.
- ACCURACY
Arabic printed text read at 93.84% character accuracy in our measurement, scored against exact ground truth.
- SPEED
VisionParse read a Arabic paragraph in 0.29 seconds on an 8-vCPU server.
- LANGUAGES
The Arabic model reads the connected script with contextual letter forms — isolated, initial, medial and final shapes — and both Arabic-Indic and European digits.
- LIMIT
Fully diacritised text (harakat) is harder than plain print: the marks are small, and blur removes them first. Our figure is for standard undiacritised print.
- ACCESS
The Arabic image to text tool requires no account, no email address and no payment card.
- RETENTION
VisionParse deletes free-tool Arabic files one hour after conversion.
Where the 93.84% character accuracy comes from.
Every figure on this page traces to a dated run on hardware we name. Nothing here is a vendor estimate.
Measured 2026-08-03
- Claim
- Arabic print with Arabic-Indic numerals read at 93.84% — errors concentrate in dots and digits, not word structure.
- Method
- VisionParse standard recognition tier, single-threaded. Full methodology.
- Sample
- A 146-character Arabic document paragraph rendered with Arabic (right-to-left, connected)-correct Noto type through Pango (real text shaping), read through the production pipeline with the ara language model, and scored character-by-character against the exact source.
- Measured
2026-08-03on Contabo Cloud VPS 8 — 8 vCPU AMD EPYC, 24 GB RAM, Ubuntu 24.04.4- Limitations
- Undiacritised modern print in Naskh style; fully diacritised or calligraphic text will score lower. A single synthetic document per variant is a small sample and is not a substitute for a corpus. These figures describe format behaviour, not general accuracy. A 500-document corpus run is scheduled and will supersede them.
- Contact
- VisionParse@senithu.lk
What this tool gets wrong
The connected script itself is not the problem. Letter joining and contextual forms — the part that looks hardest — read reliably. The 6.16% of errors clustered on dot placement and Arabic-Indic digit shapes, which degrade before anything else as image quality drops.
Arabic-Indic numerals deserve a manual check on financial documents. Digits like ٢/٣ and ٧/٨ differ by small strokes and are the likeliest characters to misread. For invoices where the numbers are the point, our paid AI extraction cross-checks digits against document arithmetic.
Three steps, no account.
Drop your Arabic image
A scan, a photo or a screenshot — JPG, PNG, WEBP, TIFF, BMP and HEIC are all accepted.
We read it in the right language
This page's converter runs Arabic recognition (ara) — not English recognition guessing at Arabic (right-to-left, connected) shapes.
Take the text
UTF-8 text appears with a confidence score. Copy it or download a .txt; the file is deleted within the hour.
What this tool accepts.
| Property | Free tool |
|---|---|
| Language model | Arabic (ara), Tesseract 5 traineddata |
| Script | Arabic (right-to-left, connected) |
| Measured accuracy | 93.84% character accuracy (2026-08-03) |
| Measured speed | 0.29 s per paragraph |
| Output | Plain text, UTF-8 |
| Maximum file size | 3 MB |
| Rate limit | 5 files per hour, per IP address |
| File retention | Deleted 1 hour after conversion |
| Processing location | Frankfurt, Germany |
When to use the API instead.
This page converts one file now. To convert files from your own software, the same engine runs over HTTP from Starter upward.
# same engine, from your code curl -X POST https://api.visionparse.app/v1/ocr \ -H "x-api-key: $VISIONPARSE_KEY" \ -F "file=@image.jpg" → 202 { "jobId": "c0df8e2c…" }
Use the API when
You convert more than a few files, convert on a schedule, or need results inside another system. Batch, webhooks and bounding boxes are API-only.
Frequently asked
Does the output come out in the right direction?
Yes. The text is returned in UTF-8 logical order, so it pastes correctly into Word, Google Docs and any right-to-left-aware editor.
Can it read Arabic-Indic numerals (٠١٢٣)?
Yes — our test paragraph deliberately included them, and they are counted in the 93.84% figure. They are also the likeliest characters to misread, so check totals on financial documents.
Does it handle diacritised (mushakkal) text?
It will read it, but expect lower accuracy than our published figure: harakat are the smallest marks on the page and the first casualties of blur or low resolution.
Can it read handwritten Arabic or old manuscripts?
Not usefully on this free tool — we measured a real historical manuscript at 43.9%, which is honest for unusable. The same page read at 95.3% through the AI model our paid extraction uses, so for manuscripts and archives the paid tier is the difference between garbage and a working transcription.
Is the Arabic image to text tool free?
Yes — 5 files a day with no account, no email and no card. A free account raises the allowance to 100 pages a month.
Do you keep my Arabic documents?
Free-tool files are deleted one hour after conversion, and no customer file is ever used to train models.