Does MistralAI-Training respect robots.txt?

Mistral says webmasters can disallow it in robots.txt.

Can I verify MistralAI-Training by IP?

No. Mistral publishes IP lists for its other crawlers but lists none for this one.

MistralAI-Training AI Crawler Guide

What is MistralAI-Training?

MistralAI-Training is one of three crawlers on Mistral's crawler documentation. The other two are MistralAI-Index, which indexes pages for Mistral search, and MistralAI-User, which fetches pages when a user asks Vibe a question.

Mistral says webmasters can disallow MistralAI-Training in robots.txt, and that this crawler is not used for search indexing or to answer live queries. The documented user agent is Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; MistralAI-Training/1.0; +https://docs.mistral.ai/robots).

Unlike the other two Mistral crawlers, the documentation lists no published IP addresses for MistralAI-Training, so a request carrying this name cannot be verified from operator data.

Relevant for AI searchMistral

Is MistralAI-Training relevant for AI search?

Yes. MistralAI-Training collects web content for Mistral's model training, not for search or live answers.

Mistral says MistralAI-Training crawls web content to help build datasets for training its generative AI models, and that it is not used for search indexing or to answer live questions. Blocking it opts you out of training use without affecting whether Mistral's assistant can find and cite you through MistralAI-Index and MistralAI-User.

How to handle MistralAI-Training

Decide whether you want your content in Mistral's training data. To opt out:

User-agent: MistralAI-Training
Disallow: /

This leaves MistralAI-Index and MistralAI-User untouched, so your pages can still be found and cited in Mistral's assistant. Since no IP ranges are published, treat log entries as unconfirmed.

Examples

A publisher disallows MistralAI-Training but allows MistralAI-Index to stay visible in Mistral search.
A site owner sees the training token in logs and checks its robots.txt for a matching group.

Frequently asked questions about MistralAI-Training

Learn about AI visibility monitoring and how Promptwatch helps your brand succeed in AI search.

No. Search indexing and live answers use `MistralAI-Index` and `MistralAI-User`, which are separate crawlers.

Related bots

MistralAI-Index

MistralAI-Index crawls and indexes web content for Mistral's search feature in Le Chat. Content it indexes is not used to train Mistral's generative models.

AI Assistant