AA-3D ↗740.24 Elo
MiniMax-M3 · AA-3D · Elo ↗
MiniMax / MODEL EVIDENCE
MiniMax M3
53 benchmarks · 67 observations · 60 sources · released 2026-06-01
Scores at a glance
53 benchmarks · large dot: MiniMax M3; small dots: closest models where reported
Harvey LAB-AA v1.1 ↗0.00%
MiniMax-M3 · Harvey LAB-AA v1.1 · all-pass ↗
GPQA Diamond ↗92.93%
MiniMax-M3 · GPQA Diamond · AA implementation ↗
3 more reported results
90.91% · MiniMax-M3 · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗
81.31% · MiniMax-M3 · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗
92.68 % · MiniMax-M3 · GPQA Diamond · Accuracy · Vals AI · GPQA Diamond ↗
IFBench ↗82.86%
MiniMax-M3 · IFBench · AA implementation ↗
CritPt v2 ↗44.57%
MiniMax-M3 · CritPt v2 · code tools ↗
ErdősBench ↗2.110 / 4
MiniMax M3 · ErdősBench · mean progress / 4 · API ↗
SimoBench ↗3.968 / 7
MiniMax-M3 · SimoBench · mean progress / 7 · As published ↗
MiniMax-M3 · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % ↗
1 more reported result
26.67% · MiniMax-M3 · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % · Epoch AI · OTIS Mock AIME 2024-2025 ↗
MiniMax-M3 · Chess Puzzles · Best score (across scorers) · % ↗
1 more reported result
7% · MiniMax-M3 · Chess Puzzles · Best score (across scorers) · % · Epoch AI · Chess Puzzles ↗
MiniMax-M3 · Mystery Game Puzzles · Best score (across scorers) · % ↗
1 more reported result
7% · MiniMax-M3 · Mystery Game Puzzles · Best score (across scorers) · % · Epoch AI · Mystery Game Puzzles ↗
SimpleBench ↗45.8%
MiniMax-M3 · SimpleBench · Score (AVG@5) · % ↗
MiniMax-M3 · APEX-Agents · Pass@1 score · % ↗
GBAEval ↗0.91%
MiniMax-M3 · GBAEval · Overall score · % ↗
CritPt ↗3.71%
MiniMax-M3 · CritPt · Accuracy · % ↗
OSWorld 2.0 ↗4.6%
MiniMax-M3 · OSWorld 2.0 · Binary accuracy · % ↗
MiniMax-M3 · Surface Evolver Bench · Mean score · % ↗
ProofBench ↗18 %
MiniMax-M3 · ProofBench · Accuracy ↗
1 more reported result
18% · MiniMax-M3 · ProofBench · Accuracy · % · Epoch AI · ProofBench ↗
FrontierCode 1.1 ↗14.7%
MiniMax-M3 · FrontierCode · Main score · % ↗
LMCA ↗33.7%
MiniMax-M3 · LMCA · Score · % ↗
DTBench ↗78.93%
MiniMax-M3 · DTBench · Accuracy · % ↗
SciCode ↗47.11%
MiniMax-M3 · SciCode · Score · % ↗
Text Arena (Coding) ↗1486.7 Elo
MiniMax-M3 · WebDev Arena · Elo ↗
Epoch Capabilities Index (ECI) ↗146.87 ECI
MiniMax-M3 · General ECI · 26 October 2026 fit ↗
Vals Finance Index ↗48.37 %
MiniMax-M3 · Vals Finance Index · Accuracy ↗
Vals Multimodal Index ↗59.97 %
MiniMax-M3 · Vals Multimodal Index · Accuracy ↗
Legal Research Bench ↗29.81 %
MiniMax-M3 · Legal Research Bench · All-Pass ↗
LegalBench ↗85.42 %
MiniMax-M3 · LegalBench · Accuracy ↗
MiniMax-M3 · Harvey's Legal Agent Benchmark v1 · Accuracy ↗
Vals Finance Agent v2.1 ↗54.34 %
MiniMax-M3 · Finance Agent · Accuracy ↗
MiniMax-M3 · Excel Modeling Benchmark · Accuracy ↗
1 more reported result
47.76 % · MiniMax-M3 · Excel Modeling Benchmark v1 · Accuracy · Vals AI · Excel Modeling Benchmark v1 ↗
Tax Agent Bench ↗20.59 %
MiniMax-M3 · Tax Agent Bench · Accuracy ↗
Vals Finance Agent v2 ↗48.27 %
MiniMax-M3 · Finance Agent v2 · Accuracy ↗
CorpFin v2 ↗68.1 %
MiniMax-M3 · CorpFin · Accuracy ↗
MortgageTax & TaxEval v2 ↗68.36 %
MiniMax-M3 · MortgageTax · Accuracy ↗
1 more reported result
72.73 % · MiniMax-M3 · TaxEval · Accuracy · Vals AI · TaxEval ↗
MedCode ↗46.29 %
MiniMax-M3 · MedCode · Accuracy ↗
MedScribe ↗87.25 %
MiniMax-M3 · MedScribe · Accuracy ↗
SAFE-Teen ↗54%
MiniMax M3 · Urgent safety · share of applicable ratings marked fail ↗
2 more reported results
49% · MiniMax M3 · Health advice · share of applicable ratings marked fail · Vals AI · SAFE-Teen ↗
53% · MiniMax M3 · Relationship boundaries · share of applicable ratings marked fail · Vals AI · SAFE-Teen ↗
MMLU-Pro ↗84.22 %
MiniMax-M3 · MMLU Pro · Accuracy ↗
MMMU-Pro ↗81.16 %
MiniMax-M3 · MMMU Pro · Accuracy ↗
MiniMax-M3 · SAGE · Accuracy ↗
Vibe Code Bench ↗47.57 %
MiniMax-M3 (OpenHands) · Vibe Code Bench · original task set ↗
1 more reported result
9.22 % · MiniMax-M3 · Vibe Code Bench · Vibe Code Bench 1–100 · Vals AI · Vibe Code Bench 1-100 ↗
Code Migration v1.1 ↗20.9 %
MiniMax-M3 · Code Migration · Accuracy ↗
1 more reported result
19.93 % · MiniMax-M3 · Code Migration v1 · Accuracy · Vals AI · Code Migration v1 ↗
Terminal-Bench 4.0 ↗1.01 %
MiniMax-M3 (Mini-SWE-agent) · Terminal-Bench 4.0 · Score ↗
SkillsBench ↗51.5 %
MiniMax-M3 (OpenHands) · SkillsBench · Score ↗
SWE-bench Multimodal ↗33.33 %
MiniMax-M3 · SWE-bench Multimodal · Accuracy ↗
MiniMax-M3 (Mini-SWE-agent) · SWE-bench Verified · Score ↗
LiveCodeBench ↗82.15 %
MiniMax-M3 · LiveCodeBench · Accuracy ↗
MiniMax-M3 · CyberBench · Accuracy ↗
Public Benefits Bench v1.1 ↗64.14 %
MiniMax-M3 · Public Benefits Bench · Accuracy ↗
1 more reported result
60.69 % · MiniMax-M3 · Public Benefits Bench v1 · Accuracy · Vals AI · Public Benefits Bench v1 ↗
DrugDiscoveryBench ↗22.80%
MiniMax M3 (mini-SWE-agent) xhigh · DrugDiscoveryBench · Score ↗
MiniMax M3 · HiL-Bench (Human-in-Loop Benchmark) · Score ↗
ALE-Bench ↗640.02 Performance points
minimax-m3 · Average contest performance ↗
Vending-Bench 2 ↗2157.77167 USD
MiniMax-M3 · End-of-year balance ↗
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
25 of 34 shared benchmark comparisons
22 of 34 shared benchmark comparisons
GPT-5.6 Sol →OpenAI
21 of 34 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 ↗
- AA-3D: Interactive 3D Experience Benchmark | Artificial Analysis ↗
- CritPt v2 Benchmark | Artificial Analysis ↗
- GPQA Diamond Benchmark | Artificial Analysis ↗
- Harvey LAB-AA v1.1 Benchmark | Artificial Analysis ↗
- IFBench Benchmark | Artificial Analysis ↗
- SciCode Benchmark | Artificial Analysis ↗
- FrontierCode Leaderboard ↗
- DTBench — Conceptual Reasoning Index ↗
- LMCA — Conceptual Reasoning Index ↗
- Ale Bench — methodology and published results ↗
- APEX-Agents ↗
- Chess Puzzles ↗
- CritPt v1 ↗
- GPQA Diamond ↗
- Mystery Game Puzzles ↗
- OTIS Mock AIME 2024-2025 ↗
- Vending Bench 2 — methodology and published results ↗
- AI benchmarks & capabilities ↗
- MiniMax-M3 — GBA Eval ↗
- DrugDiscoveryBench — evaluation methodology and results ↗
- HiL-Bench (Human-in-Loop Benchmark) — evaluation methodology and results ↗
- AI Model Benchmarks — LM Council ↗
- OSWorld 2.0: Benchmarking computer-use agents on long-horizon real-world tasks ↗
- ErdősBench: research mathematics, scoped progress and proof audits ↗
- SimoBench | Ulam ↗
- Code Migration v1 — evaluation methodology and results ↗
- Code Migration — evaluation methodology and results ↗
- CorpFin — evaluation methodology and results ↗
- CyberBench — evaluation methodology and results ↗
- Excel Modeling Benchmark v1 — evaluation methodology and results ↗
- Excel Modeling Benchmark — evaluation methodology and results ↗
- Finance Agent v2 — evaluation methodology and results ↗
- Finance Agent — evaluation methodology and results ↗
- Vals Finance Index — evaluation methodology and results ↗
- GPQA Diamond — evaluation methodology and results ↗
- Harvey's Legal Agent Benchmark v1 — evaluation methodology and results ↗
- LiveCodeBench — evaluation methodology and results ↗
- LegalBench — evaluation methodology and results ↗
- Legal Research Bench — evaluation methodology and results ↗
- MedCode — evaluation methodology and results ↗
- MedScribe — evaluation methodology and results ↗
- SAFE-Teen: Teen Mental Health & Safety | Vals AI A conversation branches after the chatbot turns down a request — evaluation methodology and results ↗
- MMLU Pro — evaluation methodology and results ↗
- MMMU Pro — evaluation methodology and results ↗
- MortgageTax — evaluation methodology and results ↗
- ProofBench — evaluation methodology and results ↗
- Public Benefits Bench — evaluation methodology and results ↗
- Public Benefits Bench v1 — evaluation methodology and results ↗
- SAGE — evaluation methodology and results ↗
- SkillsBench — evaluation methodology and results ↗
- SWE-bench Verified — evaluation methodology and results ↗
- SWE-bench Multimodal — evaluation methodology and results ↗
- Tax Agent Bench — evaluation methodology and results ↗
- TaxEval — evaluation methodology and results ↗
- Terminal-Bench 4.0 — evaluation methodology and results ↗
- Vals Multimodal Index — evaluation methodology and results ↗
- Vibe Code Bench 1-100 — evaluation methodology and results ↗
- Vibe Code Bench — evaluation methodology and results ↗
- Surface Evolver Bench ↗
Names used in source tables
MiniMax M3 · MiniMax M3 (mini-SWE-agent) xhigh · MiniMax-M3 · MiniMax-M3 (Mini-SWE-agent) · MiniMax-M3 (OpenHands) · minimax-m3