Mistral OCR 3

Mistral OCR 3 Online for Lower-Cost Markdown OCR

Mistral OCR 3 is the cost-sensitive choice for PDFs and images that need accurate text, markdown, and table extraction. Pick OCR 3 for bulk conversion, searchable archives, and everyday document parsing when you do not need OCR 4 layout metadata.

Billing note: recognition is charged by processed page, not by uploaded file. OCR 3 uses 1 credit per processed page.

OCR 3 advantage

Lower-cost markdown OCR

Site credit usage

1 credit / page

Best when

You need many pages converted with lower credit usage

Strength

Markdown, text, tables

Upload Your Document

Supports PDF, JPG, PNG and more

Checking login status...

Why use OCR 3?

Lower cost per processed page

OCR 3 uses 1 credit per processed page on this site, so it is the practical choice for bulk uploads and cost-sensitive OCR runs.

Clean document-to-markdown output

Use OCR 3 when the main goal is turning PDFs, screenshots, scans, and images into readable markdown text and table output.

Good for everyday extraction

It works well for invoices, receipts, reports, notes, and archive documents where you need text and tables more than layout coordinates.

Best when OCR 4 metadata is optional

Choose OCR 3 when bounding boxes, block labels, and confidence scores are not required by your downstream workflow.

OCR 3 vs OCR 4: which one should you use?

OCR 3 wins when lower credit usage and clean markdown matter most. OCR 4 wins when the document pipeline needs layout-aware metadata, source regions, or confidence-based review.

Use caseOCR 3OCR 4
Budget-sensitive bulk OCRBest fit: 1 credit / pageMore metadata with higher credit usage
Markdown and text extractionStrong everyday choiceAlso strong, with richer structure
Table reconstructionGood fit for markdown tablesBetter when tables need layout context
Bounding boxes, labels, confidenceNot the main reason to choose itBest fit for layout-aware workflows

When should OCR 3 be selected?

You want the lower-credit OCR option for repeated uploads.
You mainly need markdown text, searchable content, and table reconstruction.
Your workflow does not depend on page-region coordinates or block labels.
You are testing OCR quality and cost before moving heavier jobs to OCR 4.
Start with the OCR 3 upload tool