MistralAI

Configure your Mistral API key in
Settings → Model Provider after installing the plugin.

Current chat models

Catalog reviewed on 2026-10-30 against the official model cards.
Prices below are USD per million input/output tokens.

The corresponding aliases are also available. The
official alias table
currently maps to Large 3, not the Large 4 preview.
Large 4 is also available through its major-version alias.
Aliases can change behavior and pricing upstream; the plugin's capability,
context and pricing metadata only change with plugin updates. Explicit versions
remain subject to Mistral's model lifecycle.

Mistral Embed, Codestral Embed and the existing moderation integration remain
available. OCR and audio endpoints are outside this chat catalog update.

Upgrade to 0.0.16

Medium/Large and Ministral latest definitions now expose image input. The current
models above have updated context windows and pricing. The request endpoint and
API-key configuration are unchanged.

The following retired definitions have been removed from both the YAML catalog
and model ordering. Before upgrading, migrate applications and workflow nodes
that reference these IDs. Saved workflows are not automatically rewritten, and
old IDs are not silently redirected to a different model.

Replacement models can differ in pricing, output and reasoning parameters;
review node settings and test your workflows after migration.

Deprecated definitions retained for compatibility

The current official cards label , ,
, the old Ministral 2410 models, Pixtral 12B and Nemo as
Deprecated, rather than Retired. They remain in this release, after the
current models in the picker. This is not a guarantee of API availability for
an individual account. Prefer current versions for new workflows; remove these
entries when retirement is confirmed by Mistral.

Existing Magistral latest and Devstral latest definitions also remain for
compatibility. No new deprecated model versions are introduced. The legacy
Magistral option continues to map to ; this option
is not copied to the new Medium/Small models.

This plugin currently supports predefined models only. This release does not
add custom model configuration or change the inference implementation.