Mistral Small 3
Mistral's 24B workhorse: Apache-2.0, strong and efficient.
About
Mistral-Small-24B-Instruct-2501 is Mistral's 24B dense model (January 2025): strong 24B-class quality on reasoning, math, and multilingual tasks, 32K context — Apache-2.0 licensed, not gated. 63k HF downloads. It was positioned as the open alternative to GPT-4o-mini-class models: efficient enough for single-GPU-class deployment (with quantization), capable enough for serious work. Newer Mistral Small 3.x generations supersede it, but the Apache-2.0 license keeps it a safe commercial pick.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported (Mistral, Jan 2025): competitive with GPT-4o-mini-class on MMLU, HumanEval, MATH — provider-reported.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: HF downloads 63,259 (peer-normalized); released 623 days ago. Manual review: Mistral's 24B workhorse (32K, 63k dl); superseded by Small 3.x gens; crowded 20-30B slot. License: Apache-2.0.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.2/10
- Maintenance · recency, releases, issue load
- 5.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · HF downloads 63,259 (peer-normalized)
- · released 623 days ago
- · manual review: Mistral's 24B workhorse (32K, 63k dl); superseded by Small 3.x gens; crowded 20-30B slot