Ministral-3-8B-Instruct: benchmark results

Provider: Mistral.

Aggregate Rating 1569 ± 41, rank #1299 of 10489 rated models, from 8 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
WuYuEval - Expert Module - Elo1818.82Elo rating from pairwise judged comparisons87.5
NormGym - Joint Success (Seen Layout)3.3Joint Task & Norm-Compliance Success Rate (%), Seen Layout50
NormGym - Joint Success (Unseen Layouts)4Joint Task & Norm-Compliance Success Rate (%), Unseen Layout50
NormGym - Task Success (Seen Layout)49.7Task Goal Completion Score (%), Seen Layout50
NormGym - Task Success (Unseen Layouts)50Task Goal Completion Score (%), Unseen Layouts50
WuYuEval - Foundation Module - Accuracy79.74Accuracy (%)46.9
WuYuEval - Foundation Module - Macro-F182.91Macro-F1 (%)43.8
WuYuEval - Expert Module - Judge Score0.75Mean anchor-calibrated LLM-as-a-judge score (0-1)37.5

Interactive version: theaggregate.ai/model?slug=ministral-3-8b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-20.