architectures · mistralai

Ministral 3

Compact dense transformer from the Ministral line — 16 layers with grouped-query attention (24 query heads against 8 key/value heads) and 256K positions — sized for edge and embedding workloads rather than chat. Also used as the backbone of third-party embedding models, which is why it appears under publishers other than its author. Declared as model_type `ministral3`.

Verify

What this registry has checked

Quarantine. We hold the publisher’s own published hash for these weights and have recorded it against this entry. We have not downloaded and hashed the weights ourselves, so the hash is the publisher’s claim, not our measurement.
Re-check this yourself → Source → Full record →

What is recorded

Publishermistralai
OriginFrance
LicenceApache-2.0

Elsewhere in the registry

All architectures → Browse with filters → What will run on my hardware? →

Registry record as of 2026-08-14. Identity, licence and parameter count are properties of a release and do not move; tier can, and the verify link above re-checks it against the live registry rather than this page.