Model · Mistral AI · Vision
OCR 4.1
Mistral's current OCR service, powering its Document AI stack. Adds native paragraph-level bounding boxes, structural block labels, and block-level confidence scores over OCR 3.
- Modality
- Vision
- License
- Mistral Premier (Proprietary)
- Released
- July 16, 2026
- Last verified
- September 8, 2026
- Runs locally
- No
Strengths
- Native paragraph-level bounding box extraction — layout position without a second pass
- Block-level confidence scores, so only low-confidence blocks need human review
- Structural block labels for downstream document parsing
- Batch API halves the price
Weaknesses
- Double the price of OCR 3 for the same core job ($4 vs $2 per 1,000 pages; $5 vs $3 annotated)
- Proprietary — no weights and no self-hosting, unlike most of Mistral's 2026 lineup
- Mistral publishes no context window, page limit, or file-size limit for it
Try it
| Where | Type | Notes |
|---|---|---|
| Mistral La Plateforme | hosted-api | Premier API access |
Used in solutions
Version history
- OCR 4.1 Jul 2026 Current
- OCR 3 Dec 2025 Deprecated
Official sources
- Model card docs
- Announcement announcement
Change log
- — Initial entry. Released 2026-07-16 and generally available 2026-08-31 per Mistral's changelog; mistral-ocr-latest and mistral-ocr-4 now point at it. OCR 4.0 shipped nine weeks earlier and is skipped as an interim version.
Esc