AI Potluck
Contribute
Model components / Fine-tuned / chat models

Magistral

Mistral AI

Mistral's reasoning model line, launched June 2025 in two tiers: the open-weight Magistral Small, a 24B model distilled from Magistral Medium traces with RL on top, and the API-only Magistral Medium. The current Small 1.2 adds a vision encoder and reasons in a dedicated thinking block, and fits on a single consumer GPU once quantized.

Mistral's reasoning line beside devstral (coding) and codestral (completion) on the map. Openness is read on the distributed SKU, Magistral Small, since Medium is API-only; the RL pipeline is described in the Magistral paper but not released. Verified 2026-09-01 via the Magistral-Small-2509 card and the Magistral launch post.

Openness

3 high confidence
3.0
weights
open(Magistral Small 2506/2507/2509, ungated Apache-2.0 on HF)
post-training-data
described(SFT from Magistral Medium traces + RLVR, described in the paper, not released)
code
closed(no post-training pipeline released
license
Apache-2.0(OSI)

Apache-2.0 weights for every distributed Small release, ungated on the Hub, with the post-training mixture described in the Magistral paper rather than released and no post-training pipeline. Magistral Medium is API-only and so sits outside the distributed SKU set that the license reading governs. Open weights, closed recipe: 3 by the fallthrough.

  • https://huggingface.co/api/models/mistralai/Magistral-Small-2509 recorded 2026-09-01

    `gated: false`, cardData.license `apache-2.0`, base_model mistralai/Mistral-Small-3.2-24B-Instruct-2506; 12,960 downloads in the trailing 30 days.

  • https://huggingface.co/mistralai/Magistral-Small-2509/raw/main/README.md recorded 2026-09-01

    Card: "Building upon Mistral Small 3.2 (2506), with added reasoning capabilities, undergoing SFT from Magistral Medium traces and RL on top"; "Apache 2.0 License: Open license allowing usage and modification for both commercial and non-commercial purposes"; benchmark table for Medium 1.2 and Small 1.2. No mixture or pipeline release.

  • https://mistral.ai/news/magistral recorded 2026-09-01

    The Magistral launch post (published 2025-06-10), announcing the dual release — open-weight Small and API-only Medium — which is what scopes the license reading to the Small SKUs.

Adoption

2 high confidence
2.0

94,181 downloads in the trailing 30 days across the three declared Small releases (Magistral-Small-2506 80,730; Magistral-Small-2509 12,960; Magistral-Small-2507 491), which bands at level 2 (10K-100K). The first release still carries most of the reach.

Capability

4 medium confidence
4.0

A capable open-weight reasoner: the 24B Small posts AIME-class math results at the deepseek-r1 rung — competitive open reasoning, short of the closed frontier the category's 5s hold — and the tier's max includes Medium 1.2, which is stronger still but API-only. Level with deepseek-r1 at 4.

Verified 2026-09-01