# Mistral OCR

69 Hacker News commenters recommended Mistral OCR and 40 criticized it, November 2025 to November 2026. It was recommended in 12 of 13 months.

**Verdict:** People recommend Mistral OCR for accurate document parsing with tables, equations and bounding boxes, at low per-page cost. Critics report hallucinated sentences in some versions.

## Why people recommend it

- Accurate output on documents (7 comments): "The accuracy is great, bounding boxes are must have for proper grounding for building answers by LLMs." ([source](https://news.ycombinator.com/item?id=49290517)) "Converting from PDF to markdown while retaining tables correctly seems to work well for me" ([source](https://news.ycombinator.com/item?id=46632406)) "I have been quite happy with Mistral OCR for the documents I needed to process" ([source](https://news.ycombinator.com/item?id=49291086))
- Low cost per page (6 comments): "it just runs for a year without any worries about the amount of requests I make, because the cost is minuscule" ([source](https://news.ycombinator.com/item?id=48783157)) "I prefer using a European provider though so this is actually cheaper than I can find elsewhere." ([source](https://news.ycombinator.com/item?id=50300987)) "it's cheap (2c per page), fast, easy to integrate API, and sufficient for my needs" ([source](https://news.ycombinator.com/item?id=48483573))
- Fast processing (4 comments): "Results were good, and fast... but I actually got better quality output from sending it to Haiku 4.5 instead." ([source](https://news.ycombinator.com/item?id=47431708)) "it's cheap (2c per page), fast, easy to integrate API, and sufficient for my needs" ([source](https://news.ycombinator.com/item?id=48483573)) "in internal benchmarks, Mistral OCR is significantly faster than comparable APIs" ([source](https://news.ycombinator.com/item?id=49289821))
- Data stays in EU (2 comments): "I prefer using a European provider though so this is actually cheaper than I can find elsewhere." ([source](https://news.ycombinator.com/item?id=50300987)) "it hits the exact sweet spot between cost and my data staying in the EU" ([source](https://news.ycombinator.com/item?id=47422929))

## What people warn about

- Hallucinates sentences in documents (3 comments): "For the most part Mistral is actually good, but I can't have it just inventing sentences in the middle of a document." ([source](https://news.ycombinator.com/item?id=49295606)) "it caught entirely hallucinated sentences created by Mistral OCR 4.0" ([source](https://news.ycombinator.com/item?id=49291503)) "the hallucinations in 4.0 were so egregious (just completely making up new sentences in the middle of a page)" ([source](https://news.ycombinator.com/item?id=49294642))
- Claude or Haiku gave better output (3 comments): "Results were good, and fast... but I actually got better quality output from sending it to Haiku 4.5 instead." ([source](https://news.ycombinator.com/item?id=47431708)) "The quality I was getting from Mistral OCR 2 was nowhere near as good as what I could get from just sending the same files to Claude Sonnet" ([source](https://news.ycombinator.com/item?id=47422598)) "found it was better just to write a program that sent the documents to Claude Sonnet to OCR instead" ([source](https://news.ycombinator.com/item?id=47744759))
- Confusing portal and documentation (3 comments): "the Mistral documentation and dashboards are all over the place" ([source](https://news.ycombinator.com/item?id=49607449)) "the Mistral portal is honestly kind of confusing, considering that most people just want to enter CC details and get an API key" ([source](https://news.ycombinator.com/item?id=50289931)) "It's even more confusing when you want to create an API key because they are separated by product" ([source](https://news.ycombinator.com/item?id=47423485))
- Accuracy below open-source alternatives (2 comments): "I don’t know how they can make this statement with 79% accuracy rate. For any serious use case, this is an unacceptable number." ([source](https://news.ycombinator.com/item?id=46331450)) "I'm reading worse performance than many OSS offerings like Paddle, MinerU, MonkeyOCR, etc" ([source](https://news.ycombinator.com/item?id=46331109))

## Alternatives hackers recommend

- [PaddleOCR](https://hackerapproved.com/p/paddleocr.md): 19 recommend
- [GLM-OCR](https://hackerapproved.com/p/glm-ocr.md): 16 recommend
- [DeepSeek-OCR](https://hackerapproved.com/p/deepseek-ocr.md): 12 recommend
- [dots.ocr](https://hackerapproved.com/p/dots-ocr.md): 8 recommend
- [Unlimited OCR](https://hackerapproved.com/p/unlimited-ocr.md): 7 recommend

## Where to get it

- [Visit the official site](https://mistral.ai/)

## Cite as

Hacker Approved, "Mistral OCR", https://hackerapproved.com/p/mistral-ocr/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).
