Pixtral-12B-2409 — benchmark results

Provider: Mistral. Access: Open.

Unified ELO 1288 ± 36, rank #1512 of 1716 rated models, from 20 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
KOFFVQA - Recognition65Score (%)45.1
KOFFVQA - Graph and Chart Understanding56Score (%)44.4
Clembench Multimodal v1.6.528.64clemscore (self-reported)36.8
KOFFVQA - Commonsense Reasoning65.11Score (%)30.9
KOFFVQA - Korean Recognition10Score (%)30.9
KOFFVQA - Object Attributes61.67Score (%)30.9
KOFFVQA - Table Understanding23.67Score (%)27.2
KOFFVQA44.62Overall Score (%)25.9
KOFFVQA - Relationship45.67Score (%)25.3
KOFFVQA - Korean OCR10Score (%)19.1
U-MATH - Differential Calculus16.82Accuracy (%)18.2
KOFFVQA - Hallucination and Robustness40Score (%)16.7

Interactive version: aibenchmarks.dev/model?slug=pixtral-12b-2409 · How the rankings work · Data refreshed daily, snapshot 2026-07-09.