Mistral AI: OCR 4 Outperforms 72% of Competitors in Blind Tests
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
Mistral AI: OCR 4 Outperforms 72% of Rivals in Blind Tests
Mistral AI has launched OCR 4, a new model capable of reading text from documents such as PDFs, Word files, and PowerPoint presentations.
Unlike previous versions, OCR 4 does not just extract raw text. It also identifies the position of each element on the page and its role—whether it is a title, a table, an equation, or a signature. This block classification allows for automatic division of documents into meaningful sections, which is useful for integrating them into search systems or enabling AI agents to process them. The model also provides confidence scores, estimating the degree of certainty regarding each word or page it reads.
According to Mistral, the model surpasses all tested competitors in benchmarks. OCR 4 supports 170 languages and performs well even with less common languages. In a blind test with over 600 documents, independent evaluators preferred the results of OCR 4 in 72% of cases compared to competing models. The model is available via the API, Mistral Studio, and Microsoft Foundry. Its cost is $4 per 1,000 pages, or $2 in batch mode.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.