Mistral 将 OCR 4.1 推出公开预览，支持段落级提取

Mistral 在7月16日的公开预览中增加了段落级边界框、结构标签和块级置信度得分，而其 API 指南记录了页面级和单词级的置信度输出。

By · Published

Primary source: Mistral Docs

为何重要

段落级提取为工程师提供了更细粒度的文档结构，用于搜索、代理和自动化工作流。文档化的页面级和词级置信度数据可以帮助团队将不确定的结果转交给人工验证。

Illustration of Mistral OCR 4.1 processing a document, with paragraph bounding boxes, structural labels, and block-level confidence scores highlighted for human review.

Mistral put OCR 4.1 into public preview on July 16th, 2026. The model adds native paragraph-level bounding-box extraction, structural block labels and block-level confidence scores. Mistral's current OCR documentation describes page-level and word-level confidence output.…

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Reader comments

Conversation for this story loads after sign-in.