Beyond OCR… it understands Arabic documents and their structure

Drag to upload

Full language support with excellence in Arabic and obtaining data that is copyable and directly usable

Technology developed by Misraj AI

Why does Baseer surpass traditional OCR?

Baseer transcends conventional OCR technology. It's a Vision Language Model (VLM) purpose-built to intelligently process Arabic documents and transform them into professionally structured, actionable formats.

Baseer reads texts, tables, equations, and images accurately while maintaining the original formatting

Professional processing of Arabic documents

Expert in Arabic Poetry

Baseer recognizes Arabic poetry and formats it correctly, separating the first hemistich from the second hemistich with three stars (***) to preserve the poetic meter and rhythm.

Advanced Smart Text Extraction

Basir reads documents naturally and smoothly, just like a human would, while preserving the original formatting and text order.

Mathematical Equations Support

Extracts mathematical and scientific equations in professional LaTeX format, ensuring high accuracy when handling scientific and technical content.

Professional Table Processing

It converts tables into organized HTML format ready for use, making them easy to display and process in any application.

Image Location Detection

Identifies the positions of images within a document and inserts special <img> tags in the text to distinguish them from the surrounding content.

Intelligent Handling of Special Elements

Watermarks: Recognizes watermarks (Official Copy, Confidential) and wraps them in <watermark> tags.

Upload, Edit… and Save Your Documents Easily in Word & PDF

Baseer provides a complete document processing environment that allows you to upload your Arabic files and edit their content accurately, with the ability to export the final results in Word and PDF formats according to enterprise-ready standards.

Baseer surpasses leading global models in accuracy benchmarks.

We tested 'Baseer' - our specialized model trained to understand Arabic documents and images and transform them into structured, usable text - and compared it against the world's most powerful models. Baseer demonstrated exceptional superiority in text comprehension and maintaining table structure, delivering the most accurate digital outputs achievable by any AI model today. At Misraj Lab, we take pride in developing 'Baseer' as the premier vision-language model

View Research Paper

Model	WER↓ lower is better	CER↓ lower is better	BLEU↑ higher is better	CHRF↑ higher is better	TEDS↑ higher is better	MARS↑ higher is better
★ Baseer_v2	0.19	0.12	76.31	88.94	67	77.8
gemini_2.5_pro	0.37	0.31	77.92	89.55	52	70.775
Azure Document Intelligence	0.44	0.27	62.04	82.49	42	62.245
Dots_ocr	0.5	0.4	58.16	78.41	40	59.205
Nanonets_OCR2_3B	0.78	0.71	44.29	68.39	49	58.695
GPT-5	0.86	0.62	40.67	61.6	48	54.8
Qwen2_5_vl_32b	0.76	0.59	37.62	62.64	41	51.82
Qwen3_VL_8B_Instruct	0.87	0.78	32.95	54.54	49	51.77
DeepSeek_OCR	0.88	0.81	41.57	62.33	30	46.165
MISTRAL	0.49	0.42	52.44	71.81	17	44.405
Qari	0.76	0.64	38.59	64.5	21	42.75
Gemma3_12B	0.96	0.8	19.75	44.53	33	38.765

Key Features

Comprehensive Feature Set

Data Privacy and Security

Your files are processed within a secure enterprise environment, without the need for external models or services

Easy Integration (API / Model)

Seamlessly integrate with your company's systems to automate Arabic document processing and convert them into ready-to-use data, enhancing operational efficiency and reducing costs

Wide Document Support

Baseer, not trained on handwritten documents, effectively handles a range of Arabic documents, from books, magazines, and academic articles to invoices and tabular data...

The First Model for Arabic Text Extraction

The best performing model among all open-source and closed-source models

Direct Document to Markdown Conversion

Preserves formatting, tables, headings, and lists, ready for processing or archiving

Arabic OCR from Baseer — High Accuracy

Baseer provides accurate OCR results in Arabic

Text from a book

PDF - 2 pages

Example of text extracted from a document

Extracted Text

How Baseer serves different sectors in processing Arabic documents with accuracy and reliability.

Examples of Use Cases

Companies and Organizations

Companies deal daily with large volumes of invoices, reports, and various forms. Baseer enables high-accuracy reading of these Arabic documents, automatically extracts key data, and converts them into ready-to-use information within business systems.

Benefits:

Accelerate operations and reduce time wasted on manual entry
Lower operational costs and reduce human errors
Seamless integration with existing systems such as ERP and accounting systems

Government Entities

Government entities suffer from massive accumulation of paper archives and old documents. Baseer helps convert these files into organized digital archives, while maintaining privacy and enabling on-premise operation without cloud dependency.

Benefits:

Intelligent archiving of official documents
Instant search across millions of pages within seconds
Complete independence from the cloud and full data control

Analysis and Research

Baseer serves researchers and educational institutions in handling scanned or complex content, converting it into structured text that can be easily searched and analyzed.

Benefits:

Convert scanned notes and lectures into usable digital text
Archive theses, dissertations, and academic references
Efficiently analyze large PDF files and extract information from them

Specialized articles to help you maximize the use of text conversion and extraction tools

Insights and articles on text processing and conversion

Various Articles

Baseer: The Vision-Language Model from Misraj

Baseer is a Vision-Language Model (VLM) we developed at Misraj to convert Arabic document images and PDF files into structured text in Markdown format while preserving the structure.

Various Articles

Why Do Global OCR Tools Fail with Arabic Text?

We attempted to copy content from a PDF file to utilize it in a project, only to be surprised by inverted letters, bizarre symbols, and erratic spacing. In fact, the tables had completely vanished. When we resorted to an OCR tool to extract the text digitally, it succeeded in pulling the content, but tables were reduced to scattered, disjointed rows, and the word order was flipped upside down.

Various Articles

From PDF files to editable files

Arabic Optical Character Recognition (OCR) remains a notoriously difficult task. The language’s cursive script, complex ligatures, diverse fonts, right-to-left orientation, and critical use of diacritics make it a nightmare for standard models. While modern Multimodal Large Language Models (MLLMs) perform brilliantly on high-resource languages, their performance on Arabic documents is severely limited.

Discover Other Blogs

Start Digitizing Your Arabic Documents Today

A model that outperforms all globally available solutions in processing Arabic content. Simply put, Baseer understands Arabic better than any other model in its field, allowing you to upload your files, edit their content with precision, then save them in Word or PDF formats ready for use within your systems or direct sharing

Choose Your Plan

Choose the plan that fits your needs

FAQs

Frequently Asked Questions

Unlike traditional Optical Character Recognition (OCR) systems that analyze images at the component level and may produce unstructured text with incorrect content order, Baseer relies on an advanced vision-language model that understands the entire image and extracts structured and accurate text while preserving context and layout.