# MistralAI-Training > MistralAI-Training is Mistral's crawler for collecting training data. Mistral's training corpus crawler. Mistral is gathering training data. **Category:** AI crawlers **Also known as:** Mistral MistralAI-Training **Updated:** 2026-09-09 **Source:** https://truestat.io/glossary/mistral-mistralai-training --- ## What MistralAI-Training is Mistral documents three crawlers with one job each and publishes an address file for two of them. For a European operator this is close to a reference implementation of how to crawl transparently. On this site MistralAI-Training is counted as training, which is the row it appears in on your dashboard. The category matters because it decides which number moves when this crawler visits, and the four are reported separately for exactly that reason. ## What its traffic looks like Broad and slow: many pages, few repeats, spread over days or weeks rather than concentrated. Training crawls tend to want everything once, so the long tail of a site sees traffic it never gets from search — old posts, deep archive pages, anything reachable by a link. Bandwidth is the usual reason people notice these at all. ## How it is verified Checked as published ip ranges. The user agent has to carry the token, and then the address is compared against what Mistral publishes — refetched every six hours, so a newly announced range is picked up without a deploy. A request that matches neither is recorded as unverified rather than as a forgery: a published list can be incomplete, and an accusation needs evidence. Only a reverse DNS record resolving to somebody else produces a spoofed verdict, which is proof rather than the absence of it. ## If you block it Mistral stops collecting your pages as training material. Nothing about how you appear in search or in live answers changes — those are separate crawlers with separate rules. Nothing you see today changes if you block this. Training crawls feed models that ship months later, so the decision is about whether your work becomes part of a commercial model, not about this quarter's traffic. The honest counter-argument is that being in the training data is part of how a model comes to mention you at all. There is no measurement that settles this either way — anyone who tells you there is has not looked — so it is a decision about your terms rather than about your numbers. ## How it differs from Mistral's other crawlers Mistral sends 3 crawlers and gives each its own token, so they can be allowed and refused separately. The tokens differ by a few characters and what the fetch means does not: MistralAI-User — ai assistant. Le Chat fetching a page to answer someone's question. MistralAI-Index — search indexing. Builds Mistral's search index. Confusing two of these is the most common mistake with this operator, and it is expensive in one direction: a rule meant for the training crawler that lands on the search crawler removes you from results without stopping any training. ## Questions ### Is MistralAI-Training in my logs really Mistral? Not necessarily, and that is why it is worth checking. Any script can send MistralAI-Training as its user agent. Mistral publishes the addresses its crawler uses, so the claim can be tested against them — a request from outside those ranges is recorded as unverified rather than accepted. ### How do I tell MistralAI-Training apart from Mistral's other crawlers? By the token, and only by the token. Mistral also sends MistralAI-User, MistralAI-Index, and the strings differ by a few characters while the jobs differ completely — one may be answering somebody's question while another collects training material. Match on the exact token rather than on the operator's name appearing anywhere in the header, or the four end up counted as one. ### Does MistralAI-Training run JavaScript? No. It requests the HTML and leaves. That is why a browser-based analytics tag never records these visits: the tag is JavaScript, and nothing runs it. Seeing this crawler at all requires reading it server-side. ### Why is MistralAI-Training counted as Training and not something else? Because that is the job this particular crawler does. Mistral runs more than one — MistralAI-User, MistralAI-Index — and they are counted separately because the decisions about them are separate. Filing them together under one operator name would let ordinary indexing look like AI interest, which is the single most misleading thing an AI traffic report can do. ### Why does MistralAI-Training not show up in Google Analytics? Google Analytics runs in the browser. This crawler never opens a browser — it requests the HTML, reads it, and leaves, so the tracking script is never executed and no event is ever sent. Every browser-based analytics tool has the same blind spot, which is why crawler traffic has to be read from the server side to be seen at all. ### What happens if a MistralAI-Training visit cannot be verified? It is recorded as unverified, not as a forgery. A published address list can be incomplete or behind — an operator can start using a range before it appears in the file — so a miss is not evidence of anyone pretending. A visit is only marked as spoofed when the address's own reverse DNS record resolves to somebody else, which is proof rather than absence of proof. ## Related - https://truestat.io/glossary/mistral-mistralai-user - https://truestat.io/glossary/mistral-mistralai-index - https://truestat.io/glossary/fcrdns - https://truestat.io/glossary/spoofed-crawler --- From the TrueStat glossary — https://truestat.io/glossary. Privacy-first web analytics that also shows you which AI assistants are reading your site.