In short. MistralAI-Index collects web content for Mistral search, with a dedicated crawler token and published IP ranges for reviewing requests.

Definition: As of 17 October 2026, MistralAI-Index is Mistral's automated web crawler for building its search index. Mistral describes that index as supporting answers to user questions in Vibe, the assistant named in its crawler documentation.

Key facts

Mistral's crawler documentation states that content collected by MistralAI-Index is excluded from model training.
The IETF's Robots Exclusion Protocol defines robots.txt as a file of crawler access rules at the service's top-level path.
Mistral's published address file provides the crawler's network ranges in a JSON prefixes array.

What is the indexing task?

An index is the searchable collection built from crawled content. Mistral's description of MistralAI-Index assigns it automated web collection for search indexing only. Mistral search then helps answer questions in Vibe.

The same documentation assigns dataset collection for model training to MistralAI-Training. That separate crawler identity matters when defining website rules: a group naming MistralAI-Index addresses search indexing activity. A training collection policy uses the training crawler's own token, the name used to select its rules.

Which control rules apply?

A robots.txt group associates a crawler name with path rules. The following example expresses a request to exclude MistralAI-Index from all paths, using the syntax defined by the Robots Exclusion Protocol:

User-agent: MistralAI-Index
Disallow: /

The file belongs at /robots.txt, served as UTF-8 plain text. User-agent selects the crawler, and Disallow: / covers paths beginning at the root. For an explicitly permissive group, the corresponding path rule is Allow: /.

Under the protocol, matching named groups supply the rules; multiple matching groups combine. The wildcard group, User-agent: *, supplies fallback rules when no named group matches. Adding a named group therefore requires reviewing the restrictions that should apply within it.

Mistral's documentation introduces robots.txt controls but leaves Index-specific compliance details and rule update timing unspecified.

How can a request be checked?

Mistral links a dedicated MistralAI-Index IP address file. Its creationTime field records a timestamp, while each ipv4Prefix value identifies an IPv4 network range.

For a request under review, compare the source IP address with those published ranges. Keep the retrieved file with the review record so that the comparison refers to a particular copy. The crawler name and the address comparison provide separate items to inspect when reviewing traffic attributed to MistralAI-Index.

FAQ

Does allowing MistralAI-Index also permit training collection?

Mistral's crawler documentation assigns training collection to MistralAI-Training and excludes Index-collected content from that use. Review the training token separately when setting your collection policy.

Can robots.txt protect confidential pages?

Use access controls for confidential material. The Robots Exclusion Protocol identifies HTTP authentication as a security measure and warns that paths listed in robots.txt are public.

Sources

Check how a page reads to answer engines. Open the GEO score analyzer