Magistral Small 1.2

Mistral AIMULTIMODAL Open weight
Provider
Mistral AI
Modality
MULTIMODAL
Parameters
24B
Context window
128,000 tokens
Weights
Open
Released
18 Sept 2025
Compare Magistral Small 1.2 with others

Magistral Small 1.2 (Sept 2025) is a 24B Apache-2.0 reasoning model with strong AIME math and transparent chain-of-thought that runs on a single consumer GPU.

Best for

  • Multi-step reasoning and AIME-class math with visible thinking
  • Competitive coding (LiveCodeBench ~71)
  • Multimodal reasoning (added a vision encoder in 1.2)
  • An open reasoning model you fully control

How it compares — and the India angle

  • Standout self-host reasoning option for India: runs on a single RTX 4090 or 32GB Mac at 4-bit (~15-19GB).
  • Zero USD billing when self-hosted, unlike API-only reasoning models.
  • Magistral Medium 1.2 (proprietary API) scores higher for those who prefer managed access.

Benchmarks

Representative public scores (approximate, higher is better) for relative comparison. Check the provider for the latest official results.

GPQA Diamond70
AIME (math)86

How to access

API ~$0.50 / 1M input, ~$1.50 / 1M output. Apache 2.0 (free to self-host).

Access Magistral Small 1.2
Share: