Mistral Large 4: benchmark results

The open-weight release of Mistral Large 4 on October 29, 2026, which Mistral describes as an improved model after further reinforcement learning on the October 6 public preview. Natively multimodal MoE with about 1T total and 52B active parameters; Mistral publishes FP8 weights, with an NVFP4 quantization in a separate repository. Provider: Mistral. Released 2026-10-29. Access: Open.

Aggregate Rating 1768 ± 8, rank #144 of 10516 rated models, from 325 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AI Coding Daily (OpenCode) - React-TS Code Quality18.5React-TS Code Quality (max 20) points, LLM-judged rubric sco100
Mistral Large 4 Release - B3 Agent Security - Attack Resistance97.1Attack Resistance (%)100
Mistral Large 4 Release - Cognition FrontierCode 2.019.7Score (%)100
Mistral Large 4 Release - Cyber Refusal (JailbreakBench, StrongREJECT, AgentHarm)96.6Average Refusal Rate (%)100
AA GDP.pdf - Legal - Criterion Pass Rate88.25Criterion Pass Rate (%)99.6
Nejumi 4 - MT-Bench (Japanese) - Writing99.5Judge rating (1-10, x10)97.2
OTIS Mock AIME 2024-2599.44Accuracy (%)92.9
AA-Omniscience Hallucination Rate29.81Hallucination Rate (%)92.6
Nejumi 4 - Toxicity - Violation Categories54.76Criteria met (%)91.6
AA-Briefcase - DOCX - Rubric Pass Rate57.4Rubric Pass Rate (%)91.3
SecureBio BioTIER Permit99.9Benign prompts answered (%)90.8
Nejumi 4 - MT-Bench (Japanese) - Coding100Judge rating (1-10, x10)90.1

Interactive version: theaggregate.ai/model?slug=mistral-large-4 · How It Works · Data refreshed daily, snapshot 2026-11-21.