CritPt ↗0%
Mistral Large 3 · CritPt · Accuracy · % ↗
Mistral / MODEL EVIDENCE
Mistral Large 3
20 benchmarks · 21 observations · 21 sources · released 2025-12-02
Scores at a glance
20 benchmarks · large dot: Mistral Large 3; small dots: closest models where reported
LMCA ↗16.7%
Mistral Large 3 · LMCA · Score · % ↗
DTBench ↗65.07%
Mistral Large 3 · DTBench · Accuracy · % ↗
Mistral Large 3 · GDP.pdf · GDP.pdf score · % ↗
SciCode ↗36.57%
Mistral Large 3 · SciCode · Score · % ↗
Text Arena (Coding) ↗1229.52 Elo
Mistral Large 3 · WebDev Arena · Elo ↗
LegalBench ↗79.14 %
Mistral Large 3 · LegalBench · Accuracy ↗
CaseLaw v2 ↗61.41 %
Mistral Large 3 · CaseLaw · Accuracy ↗
CorpFin v2 ↗61.03 %
Mistral Large 3 · CorpFin · Accuracy ↗
MortgageTax & TaxEval v2 ↗52.11 %
Mistral Large 3 · MortgageTax · Accuracy ↗
1 more reported result
73.06 % · Mistral Large 3 · TaxEval · Accuracy · Vals AI · TaxEval ↗
MedQA (USMLE) ↗82.23 %
Mistral Large 3 · MedQA · Accuracy ↗
AIME 2024–2025 · Vals ↗42.92 %
Mistral Large 3 · AIME · Accuracy ↗
MGSM (Multilingual GSM) ↗85.42 %
Mistral Large 3 · MGSM · Accuracy ↗
GPQA Diamond ↗68.43 %
Mistral Large 3 · GPQA Diamond · Accuracy ↗
MMLU-Pro ↗79.82 %
Mistral Large 3 · MMLU Pro · Accuracy ↗
MMMU-Pro ↗66.19 %
Mistral Large 3 · MMMU Pro · Accuracy ↗
Mistral Large 3 · SAGE · Accuracy ↗
SWE-bench Verified ↗41.4 %
Mistral Large 3 (Mini-SWE-agent) · SWE-bench Verified · Score ↗
LiveCodeBench ↗55.34 %
Mistral Large 3 · LiveCodeBench · Accuracy ↗
DistressBench ↗58.77%
mistral-large-3 · DistressBench · Score ↗
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
11 of 14 shared benchmark comparisons
11 of 14 shared benchmark comparisons
GPT-4.1 Mini →OpenAI
10 of 14 shared benchmark comparisons
How similarity is measured
Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.
Sources
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
- Arena leaderboard — WebDev evaluation ↗
- SciCode Benchmark | Artificial Analysis ↗
- DTBench — Conceptual Reasoning Index ↗
- LMCA — Conceptual Reasoning Index ↗
- CritPt v1 ↗
- DistressBench — evaluation methodology and results ↗
- GDP.pdf Benchmark | Surge AI ↗
- AIME — evaluation methodology and results ↗
- CaseLaw — evaluation methodology and results ↗
- CorpFin — evaluation methodology and results ↗
- GPQA Diamond — evaluation methodology and results ↗
- LiveCodeBench — evaluation methodology and results ↗
- LegalBench — evaluation methodology and results ↗
- MedQA — evaluation methodology and results ↗
- MGSM — evaluation methodology and results ↗
- MMLU Pro — evaluation methodology and results ↗
- MMMU Pro — evaluation methodology and results ↗
- MortgageTax — evaluation methodology and results ↗
- SAGE — evaluation methodology and results ↗
- SWE-bench Verified — evaluation methodology and results ↗
- TaxEval — evaluation methodology and results ↗
Names used in source tables
Mistral Large 3 · Mistral Large 3 (Mini-SWE-agent) · mistral-large-3