Mistral OCR 4 Launches With Enhanced Document Processing Features

WHO:

MarkTechPost - Mistral AI is the organization that released OCR 4.

WHAT:

MarkTechPost - Mistral OCR 4 features structured document output, moving from clean text extraction.

Techmeme - The model includes bounding boxes, block classification, and inline confidence scores.

MarkTechPost - The model supports 170 languages.

WHERE:

MarkTechPost - The model runs in a single self-hosted container.

WHEN:

MarkTechPost - Mistral OCR 4 was released on June 23, 2026.

HOW:

MarkTechPost - The model feeds citation-ready inputs into RAG, agentic, and enterprise search pipelines through one API endpoint.

(UPDATED June 24, 2026 01:37 UTC)

Mistral AI has just released its latest innovation, Mistral OCR 4, which features structured document output that goes beyond simple text extraction. This new model supports 170 languages and includes advanced functionalities like bounding boxes, block classification, and inline confidence scores, all running in a single self-hosted container. With its ability to feed citation-ready inputs into various search pipelines through one API endpoint, Mistral OCR 4 is set to enhance document processing capabilities significantly.