xSeek/Docs/Mistral AI
Back to documentation

Mistral AI user agents, explained.

Mistral ships a single user-initiated crawler for LeChat. No training crawler, no silent indexing — MistralAI-User only visits when a real user asks.

Mistral AI · OfficialUpdated Apr 2026~5 min read1 user agent
TL;DR

MistralAI-User/1.0 is Mistral's only documented web fetcher. It fires when LeChat users ask a question that requires web content. Verify via mistral.ai/mistralai-user-ips.json.

Overview

Mistral AI is a Paris-based company building open-weight language models and LeChat, their consumer assistant. Unlike GPTBot or ClaudeBot, Mistral doesn't ship a training crawler for the public web.

MistralAI-User only visits your site when a real LeChat user asks a question that requires fetching a URL — and when it does, it cites you in the response.

How to identify the agents

1 crawler

Mistral AI identifies itself with specific user-agent strings. Verify each through a published IP range.

MU

MistralAI-User

LeChat user fetches · on-demand
User-triggered
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; MistralAI-User/1.0; +https://docs.mistral.ai/robots)

Fired when a LeChat user asks a question that requires fetching a specific page. Not a training crawler, not a silent indexer — it only fires for real user requests and includes a citation to your site in the response.

Verifiablemistralai-user-ips.json

How Mistral AI accesses your content

3 access patterns, mapped to the agents above.

1

On-demand only

MistralAI-User only visits when a LeChat user asks a question that requires the page.
2

Cited responses

Every fetch results in LeChat citing your content with a link back — real referral traffic.
3

Verifiable

Mistral publishes their IP list at mistral.ai/mistralai-user-ips.json for verification.
!
No training crawler. Mistral does not operate a public training crawler. MistralAI-User is strictly on-demand and is not used to collect training data.

Control Mistral AI's access

robots.txt

You control each agent independently. Common configurations below.

Allow MistralAI-UserRecommended

Let LeChat cite you. No training risk.
# Allow LeChat to cite your content User-agent: MistralAI-User Allow: /

Block MistralAI-UserRestrictive

Removes your site from LeChat answers.
# Block Mistral AI completely User-agent: MistralAI-User Disallow: /

Block private sections only

Keep public content citable, protect private paths.
# Allow public pages, block private User-agent: MistralAI-User Disallow: /private/ Disallow: /members-only/ Allow: /

Optimize content for Mistral AI.

LeChat cites sources that answer the question directly. Optimize accordingly.

Lead with the direct answer. The first paragraph should resolve the question — LeChat often quotes the top of the page.
Server-render your content. MistralAI-User fetches pages on demand — if your answer is behind JavaScript, it won't make the citation.
Structure with semantic HTML. Headings, lists, and tables for clean extraction.
Verify IPs against the published list. Use mistral.ai/mistralai-user-ips.json server-side to confirm real Mistral traffic.
Write European-friendly copy. LeChat has strong French and European audiences — localized content lifts citation rates.

Track Mistral AI visits with xSeek.

See every MistralAI-User visit.

Monitor when LeChat users pull your content into answers. Track which pages get cited and what queries trigger the fetch.

Start free →

Frequently asked questions

Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; MistralAI-User/1.0; +https://docs.mistral.ai/robots). It's the only user agent Mistral ships for web fetches.
Source: Mistral AI's official robots documentation.
PreviousPerplexity user agentsNextBing AI user agents