Mistral OCR 4.1: the Document AI surface that gets competitive attention in the 8/13-8/14 frontier pricing cycle

2 sources

Mistral released Mistral OCR 4.1 (model identifier: mistral-ocr-4-1) to Public Preview at the Premier tier on 2026-07-16. Capabilities: native paragraph-level bounding box extraction, structural block labels, block-level confidence scores. Pricing: €3.50 per 1,000 pages; €4.38 per 1,000 annotated pages. Features available: BBox extraction, OCR, structured annotations, batching. The release predates the captured date by a month, but Document AI surfaces back into creator attention when the broader frontier pricing cycle (Gemini 3.7 Flash two-tier pricing on 2026-08-13, GPT-5.6 Sol Ultrafast mode on 2026-08-13) forces a re-evaluation of which Document AI surface fits each creator's price-per-page budget. The takeaway for creators: Mistral OCR 4.1 is a Document AI surface worth comparing against the Document AI surfaces bundled inside Gemini 3.7 Flash and any Anthropic / OpenAI Document AI tier, on a per-page and per-annotated-page basis.

Why now

The story is the editorial frame for 'Document AI surfaces in the 8/13-8/14 frontier pricing cycle' — useful because creators re-evaluating Document AI budgets this week can anchor Mistral OCR 4.1 alongside Gemini 3.7 Flash's pricing structure.

Why it is worth publishing

Demo potential: a side-by-side of Mistral OCR 4.1 vs Gemini Document AI (bundled inside 3.7 Flash) vs Anthropic / OpenAI Document AI tier on a representative document-automation workflow, measuring per-page cost and annotation reliability.

Evidence basis

Mistral docs page + HN coverage as the broader 8/13-8/14 Document AI news cycle

“Mistral OCR 4.1 has been in Public Preview since July, but it surfaces again in this week's frontier pricing cycle as the Document AI side of the comparison set.”

Angle

Frame Mistral OCR 4.1 as the Document AI surface that gets competitive attention this week (the 7/16 release predates today by a month, but Document AI surfaces back into creator attention when the broader frontier pricing cycle forces a re-evaluation).

Format

图文卡片 / Carousel

Demo idea

A 4-card carousel: card 1 'Document AI is part of this week's pricing cycle', card 2 'Mistral OCR 4.1 (€3.50 per 1,000 pages, €4.38 per 1,000 annotated pages)', card 3 'Gemini Document AI bundled inside Gemini 3.7 Flash (per-1M-token pricing)', card 4 'decision rule: per-page vs per-token budget for document automation'.

Platform notes

Anchor freshness to the 2026-07-16 release date; do not present as today's release. Per-page and per-annotated-page pricing were extracted from the Mistral docs page; re-extract from the docs before stating any specific number.

Usable claims

Mistral released Mistral OCR 4.1 (model identifier: mistral-ocr-4-1) to Public Preview at the Premier tier on 2026-07-16. Capabilities: native paragraph-level bounding box extraction, structural block labels, block-level confidence scores. Pricing: €3.50 per 1,000 pages; €4.38 per 1,000 annotated pages.

Evidence pipeline

From the news

Mistral OCR 4.1: Public Preview at Premier tier (July 2026 release)

Breakdown

Mistral OCR 4.1 was released 2026-07-16 to Public Preview at the Premier tier — useful to anchor as the Document AI surface that gets competitive attention this week but easy to misread as a fresh release on 2026-08-14. This explainer uses the per-page / per-annotated-page pricing as the lens for comparing across the Document AI surface set (Mistral OCR 4.1, Gemini Document AI bundled inside Gemini 3.7 Flash, Anthropic / OpenAI Document AI tier), while keeping the freshness anchor honest (the 7/16 release date, not 8/14).

Sources

Risks

Anchor the topic to the 2026-07-16 release date and frame it as 'the Document AI surface that gets competitive attention this week' rather than 'today's release'.

Demo ideas

Decision card: 'Mistral OCR 4.1 vs Gemini Document AI vs Anthropic / OpenAI Document AI' — pick by cost-per-page vs cost-per-token.
Workflow comparison: same document automation workflow (e.g., invoice extraction) across the Document AI surface set.
Annotation-reliability walkthrough once the per-annotated-page reliability numbers are re-extracted from Mistral docs.