Open-Weight Model Families · 2025
Magistral
Mistral's first reasoning models, trained with a from-scratch RLVR pipeline (no distillation) and documenting a full reproducible recipe; the 24B Magistral Small is open-weight under Apache 2.0.
Editorial record
Plain-language summary
Mistral's first reasoning models, trained with a from-scratch RLVR pipeline (no distillation) and documenting a full reproducible recipe; the 24B Magistral Small is open-weight under Apache 2.0.
Knowledge graph
Relationships
Descendants
Depends onEvidence: Strongly supported
Attention Is All You Need
Builds on the open weight family lineage in the archive
freshness sweep 2026
Source record
Provenance
- Record ID
- P-632
- Record created
- 2026-07-13
- Last reviewed
- 2026-07-14
- Record version
- 2
- https://arxiv.org/abs/2506.10910
- arXiv:2506.10910
Citation caveat: Citation metadata is approximate and marked unverified in the source dataset.