Mistral OCR

Name: mistral-ocr
Rating: 72
Author: pascalandy

OCR conversion using Mistral AI API. Excellent for scanned documents, images with text, and complex layouts.

Prérequis

•Copy .env.example to .env
•Add your MISTRAL_API_KEY to .env

Usage

bash

# From the mistral-ocr directory
cd .opencode/skill/convert-to-md/converters/mistral-ocr

# Basic usage
uv run scripts/mistral_ocr.py input.pdf -o output_mistral.md

# With options
uv run scripts/mistral_ocr.py input.pdf -o output_mistral.md --table-format html
uv run scripts/mistral_ocr.py input.pdf -o output_mistral.md --no-images

Options

Option	Description
`-o, --output`	Output file path (default: `{input}_mistral.md`)
`--table-format`	`markdown` (default) or `html`
`--no-images`	Exclude base64 images from output

Supported Formats

•PDF (.pdf)
•Images: PNG, JPG, JPEG, GIF, WEBP

Notes

•Best results for scanned documents and images
•Preserves table structure
•Can extract text from images within PDFs
•API usage incurs costs based on Mistral pricing