
Mistral Small 3.2
The 24B open workhorse refresh - better instruction following, 128K context, Apache 2.0.
About
Mistral Small 3.2 (25.06) is Mistral's June 20, 2025 refresh of the popular 24B Small line. It keeps the same 23.6B-parameter open-weight architecture but fixes the pain points of 3.1: better instruction following, fewer infinite-generation and repetition errors on hard prompts, improved tool-calling robustness, and a more refined tone.
It is multimodal (text and image input), carries a 128K context window, and remains one of the most-downloaded open-weight families on Hugging Face - the default choice for on-prem chatbots, RAG, and agent workloads that need Mistral quality at self-hostable scale.
Honest limits: it is not a reasoning model - for chain-of-thought workloads, Mistral's Magistral line (or Medium 3.5's reasoning mode) is the intended path. Knowledge cutoff is October 2023.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 88; Trending score 80/100 from researcher assessment.