Mistral Small 3.1 24B
Open Sourcevon Mistral AI · Veroeffentlicht 2025-03-17
55.8
Durchschn. Score
$0.35/1M
Eingabepreis
$0.56/1M
Ausgabepreis
128K tokens (~64 books)
Kontextfenster
multimodal
Typ
Tested on 5 benchmarks with 55.8% average. Top scores: HELM — WildBench (78.8%), HELM — IFEval (75.0%), HELM — MMLU-Pro (61.0%).
Benchmark-Ergebnisse
| Benchmark | Kategorie | Score | Bar |
|---|---|---|---|
| HELM — WildBench | reasoning | 78.8 | |
| HELM — IFEval | language | 75.0 | |
| HELM — MMLU-Pro | knowledge | 61.0 | |
| HELM — GPQA | knowledge | 39.2 | |
| HELM — Omni-MATH | math | 24.8 |
Aehnliche Modelle
Alibaba Qwen
55.9
OpenAI
56.0
DeepSeek
56.0
OpenAI
56.2