Mistral Medium 3.5 (High): benchmark results

Provider: Mistral. Released 2026-04-29. Access: Open.

Unified ELO 1544 ± 1, rank #656 of 1761 rated models, from 35 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
PACT (Lechmazur)1538PACT Bilateral Rating60
GRIPS79.2Accuracy (%)46.2
GRIPS-hard26.2Accuracy (%)46.2
Vals AI CorpFin v258.78Accuracy (%)37.6
Vals AI Finance Agent46.11Accuracy (%)36.7
Vals AI Terminal-Bench 2.030.34Accuracy (%)34.1
Vals AI SAGE37.61Accuracy (%)32.9
Vals AI Harvey Legal Agent Bench0.42Accuracy (%)28.1
Vals AI TaxEval v267.99Accuracy (%)27.8
ProofBench10Score (self-reported)27.2
MedCode33.75Score (self-reported)23
Vals AI MedCode33.75Accuracy (%)21.3

Interactive version: theaggregate.ai/model?slug=mistral-medium-3-5-high · How It Works · Data refreshed daily, snapshot 2026-09-05.