Bots
MistralAI-User: Mistral's live fetcher
MistralAI-User is the agent Mistral uses when someone asks its assistant a question that needs a page. Mistral documents that the response links to the source, and that this fetcher neither crawls automatically nor collects training data. Allow it if you want to be cited. Mistral publishes its addresses, so visits can be verified.
7 days free · card required
0requests filed as MistralAI-User · 30 days
of 143,601 AI crawler requests from 7 operators
| Agent | AI requests, 30 days | attensira.com robots.txt |
|---|---|---|
| MistralAI-User (this page) | 0 | Allowed |
| ChatGPT-User | ≈230 | Allowed |
| Claude-User | ≈36 | Allowed |
| Amzn-User | 0 | Wildcard |
| Perplexity-User | 0 | Allowed |
2 of 5 had requests filed in the window. User agents as sent, not checked against address lists.
Analyst · reads the AI traffic boardAI traffic board, read over the Attensira MCP, and our live robots.txt · · n=143,601 AI requests
Karl-Gustav Kallasmaa, Founder & CEOLast updated How do I allow or block MistralAI-User in robots.txt?
Paste this into the robots.txt at the root of your domain. Beside it, the group attensira.com serves MistralAI-User today.
# Let Mistral read your pages when someone asks about you.User-agent: MistralAI-UserAllow: /User-agent: MistralAI-UserAllow: /Disallow: /api/We name MistralAI-User and allow it everywhere except /api/, the same as every AI and search crawler we allow by name.
Read the whole file, or write yours with the robots.txt generator.
Who runs MistralAI-User, and what is it for?
Mistral AI runs it, and we file it under user-triggered fetch. Fetches a page when someone asks Mistral's assistant a question, and the response includes a link to the source.
From Mistral AI’s documentation, read . Why a training fetch and a search fetch are different decisions: the AI crawler guide.
Does MistralAI-User obey robots.txt?
Every other user-triggered fetcher in this directory carries an exemption in the operator's own words. OpenAI: robots.txt rules may not apply, because the fetch follows a user action. Perplexity: "since a user requested the fetch, this fetcher generally ignores robots.txt rules". Amazon: "may not follow all robots.txt directives". Meta: "may bypass robots.txt rules".
Mistral claims no exemption. Its wording is that "MistralAI-User governs which sites these user requests can be made to", on a page framed around robots.txt tags.2 This site records the policy as respects, a deliberate departure from the other user-triggered pages. It is a reading of an absence rather than an explicit compliance statement, flagged so the next person to re-fetch Mistral's page knows what to check.
The practical result is that a rule here is likely to be a real control, so write it with care. Name the token in its own group rather than relying on the wildcard: under RFC 9309, once MistralAI-User has a group, nothing in User-agent: * reaches it. The full set of blocks is in the robots.txt chapter, and our generator writes all three Mistral tokens as separate groups.
What does a MistralAI-User fetch tell you?
A verified MistralAI-User fetch will not appear in your analytics as a session and may never produce a click. It is still the earliest observable evidence that you are in an answer path, earlier than referral traffic and much earlier than a rank tracker. Because Mistral documents a source link in the response, it is one of the few tokens where the fetch and the citation are documented as connected.
Keep it in its own row. Verified user fetches show live demand, people asking Mistral things your pages answer, while MistralAI-Index fetches are routine maintenance. Averaged together they say nothing. For the same reason, never filter on the MistralAI prefix, which also covers MistralAI-Training and erases the line between training and answering.
How do I verify that a visit really came from MistralAI-User?
Check the source address against mistral.ai/mistralai-user-ips.json. Mistral publishes it separately from the index file, so a verified request tells you which job it was regardless of the header. When the header and the address disagree, believe the address. Verify before counting or exempting from rate limits: a token that sites let through freely is exactly the header worth forging. The reasoning is in verifying a crawler.
- IP range file
- mistralai-user-ips.json
Mistral AI publishes no reverse-DNS suffix for MistralAI-User; verify by the range file.
ResearcherTranscribed from Mistral AI documentation · How often do sites block MistralAI-User?
Measured by us, not reported by Mistral AI: the robots.txt of the 10,000 highest-ranked domains on the Tranco list for 2026-09-01. A domain that never mentions the token wrote no policy for it.
The method and every row, with the CSV.
- disallow MistralAI-User at the root
- 5.4%
- 434 of 8,036 domains
- allow it by name
- 0.5%
- 41 domains
- never mention it
- 94.1%
- 7,561 domains
What user-agent string does MistralAI-User send?
Match the token, never the version. A rule pinned to a version stops matching silently the next time the operator ships.
The version in this string changes; check Mistral AI’s page before relying on the exact text.
What does allowing MistralAI-User cost?
Allowing it costs one fetch each time someone's question needs your page, and that answer includes a link back to you. Mistral says it does not crawl on its own or collect training data, so blocking it gives up citations and protects nothing.
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; MistralAI-User/1.0; +https:/Should I allow MistralAI-User in my situation?
Concrete cases, and what each one implies for this agent.
You are deciding which Mistral tokens to allow and want to keep the citation path open.
Allow MistralAI-User and MistralAI-Index. Disallow MistralAI-Training if training is your objection.
Mistral documents the user fetch as including a link to the source in the response, and states that neither this token nor the index crawler is used for generative AI training. The citation path and the training path are entirely separate tokens here, so no trade-off is required.
You blocked every user-triggered fetcher across all operators on principle.
Reconsider this one specifically, because Mistral's documented behaviour differs from the others.
OpenAI, Perplexity, Amazon and Meta each document their user-triggered fetcher as not bound by robots.txt. Mistral instead says the token governs which sites these user requests can be made to. Blocking here plausibly works, which means it plausibly costs you the citation the others would have given you anyway.
A burst of MistralAI-User requests follows a launch and you suspect scraping.
Verify against the published address file before rate-limiting.
A token attached to live user questions is worth impersonating, precisely because sites exempt it from rate limits. If the addresses are not in Mistral's file, you are throttling something else entirely — and if they are, you are throttling people asking about your launch.
What else do people ask?
What is MistralAI-User?
Does MistralAI-User crawl my site on its own?
Does it collect training data?
Where do these facts come from?
5 claims on this page, each with the page it was read from; last checked . The numbers in the text point here.
Mistral states that MistralAI-User is for user actions in its assistant, and that when users ask a question it may visit a web page to help answer and include a link to the source in its response.
docs.mistral.ai · retrieved
“When users ask Vibe a question, it may visit a web page to help answer and include a link to the source in its response.”
Mistral states that MistralAI-User governs which sites these user requests can be made to, and that Mistral uses specific robots.txt tags to help webmasters manage how their sites and content interact with AI. Mistral's page does not state that this fetcher ignores or may bypass robots.txt.
docs.mistral.ai · retrieved
“MistralAI-User governs which sites these user requests can be made to.”
Mistral states that MistralAI-User is not used for crawling the web in any automatic fashion, nor to crawl content for generative AI training, and separately that content crawled by MistralAI-Index is not used for generative AI training of any kind.
docs.mistral.ai · retrieved
“It is not used for crawling the web in any automatic fashion, nor to crawl content for generative AI training.”
Mistral publishes MistralAI-User's full user-agent string as "Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; MistralAI-User/1.0; +https://docs.mistral.ai/robots)".
docs.mistral.ai · retrieved · changes often, check the source
Mistral publishes the addresses MistralAI-User fetches from at mistral.ai/mistralai-user-ips.json, separately from the MistralAI-Index address file.
docs.mistral.ai · retrieved
Want whatever it takes to win the answer?
7 days free · card required