UnifyBench

RankingEffort benchmarks

MiniMax-M3

Compare models

44 measured results · 3 effort settings · 40 benchmark/harness combinations

Model evidence and sources →

Change model

Compare with

Choose a second model and its effort to see both profiles and benchmark differences below.

Compare with

Capability profile

One configuration, seven capabilities

Comparison score · 0–100
MiniMax-M3 · Reasoning capability radarComparison scores from 0 to 100. Unknown capabilities have no point; lines stop at gaps. Hollow points are preliminary. Exact values and support labels follow the chart.AgenticHard reasoningCodingHuman prefKnowledgeMultimodalLong context50100MiniMax-M3 · Reasoning · Agentic: 24.5 · SupportedMiniMax-M3 · Reasoning · Hard reasoning: 37.4 · SupportedMiniMax-M3 · Reasoning · Coding: 32.7 · PreliminaryMiniMax-M3 · Reasoning · Knowledge: 14.0 · SupportedMiniMax-M3 · Reasoning · Multimodal: 36.7 · PreliminaryMiniMax-M3 · Reasoning · Long context: 58.1 · Preliminary

Gaps are unknown, not zero. Hollow points are preliminary.

  • Agentic

    A24.5Supported

  • Hard reasoning

    A37.4Supported

  • Coding

    A32.7Preliminary

  • Human pref

    Unknown

  • Knowledge

    A14.0Supported

  • Multimodal

    A36.7Preliminary

  • Long context

    A58.1Preliminary

Each profile belongs to the named configuration. Capability scores use a shared panel for all axes, distinct from the overall ranking panel; they are not accuracy percentages or direct head-to-head wins.

Capability score evidence

MiniMax-M3 · Reasoning: 16.9 overall comparison score. Meets the overall evidence requirements.

6 areas measured. The available task mix affects this estimate; missing areas remain unknown.

Which sources carry the weight?
  • Artificial Analysis · 100.0% direct comparison weight · independent of the model provider
Which benchmark families carry the weight?
  • IFBench · 6.9% direct comparison weight
  • Humanity’s Last Exam · 6.8% direct comparison weight
  • Mmmu Pro · 6.8% direct comparison weight
  • Critpt · 6.8% direct comparison weight
  • GPQA · 6.8% direct comparison weight
  • Other families (12) · 66.0% direct comparison weight

These shares describe direct comparison weight, not fractions of the final score or confidence. Source balancing does not give every family equal weight or remove correlations between overlapping sources.

How sensitive is this score?

The audit removes one source at a time and refits against the same overall reference panel. These published results use the matching dataset audited 2026-10-23.

Both ranks use the same surviving configurations for each removal. The cohort starts with 444 configurations supported in the original audit, including separate effort settings. These are diagnostic positions, not the default leaderboard ranks. Eligibility is not rechecked after removal.

What changes when a source is removed?
Removed sourceRank before → afterRank movementCohort remaining / lost supportScore after (change)
Artificial AnalysisComparison support lost; no rank203 remaining / 241 lostUnavailable
ARC Prize156 → 155Up 1444 remaining / 0 lost20.8 (+3.9 points)
Datacurve156 → 158Down 2444 remaining / 0 lost13.9 (-3.1 points)
LiveCodeBench156 → 156No movement444 remaining / 0 lost16.8 (-0.1 points)
Scale156 → 155Up 1444 remaining / 0 lost18.9 (+2.0 points)
Terminal-Bench156 → 156No movement444 remaining / 0 lost16.8 (-0.1 points)
Cognition156 → 160Down 4444 remaining / 0 lost16.2 (-0.8 points)
LiveBench155 → 154Up 1443 remaining / 1 lost17.5 (+0.6 points)
SWE-rebench156 → 155Up 1444 remaining / 0 lost17.3 (+0.3 points)
Epoch AI152 → 155Down 3431 remaining / 13 lost17.4 (+0.5 points)
Vals AI156 → 154Up 2443 remaining / 1 lost16.3 (-0.6 points)
Provider reports · Anthropic156 → 156No movement444 remaining / 0 lost16.9 (-0.0 points)
Provider reports · Moonshot AI156 → 156No movement444 remaining / 0 lost16.5 (-0.4 points)
Provider reports · Meta156 → 156No movement444 remaining / 0 lost16.9 (-0.0 points)
Provider reports · Mistral156 → 156No movement444 remaining / 0 lost16.9 (+0.0 points)
Provider reports · OpenAI156 → 156No movement444 remaining / 0 lost17.0 (+0.1 points)
All provider reports156 → 156No movement444 remaining / 0 lost16.6 (-0.4 points)

Lost support means a configuration no longer has a comparison path to the complete reference panel. Its rank is unavailable. A surviving path alone does not establish leaderboard eligibility.

Overlapping sources may remain correlated even with source balancing. Leave-one-source-out movement is a sensitivity diagnostic, not a spread of source-specific ranks or a confidence measure. Scoring methodology · Dated configuration audit

Every benchmark family
Removed evidenceRecomputed scoreChange
omniscience20.3+3.4
arc-agi20.2+3.3
deepswe13.9-3.1
aa-lcr15.6-1.3
gpqa15.6-1.3
ifbench15.6-1.3
analyst-agent17.9+0.9
frontiercode16.2-0.8
briefcase16.2-0.7
simpleqa17.6+0.7
epoch-game-puzzles16.3-0.7
gdpval16.3-0.7
terminal-bench-science17.6+0.6
enterprise-ops17.5+0.6
gdp-pdf17.5+0.5
mmmu-pro17.4+0.4
mmlu-pro17.3+0.4
hle17.3+0.4
terminal-bench16.6-0.4
swe-rebench17.3+0.3
livebench-instructions17.2+0.3
multi-swe-bench16.6-0.3
critpt17.2+0.2
livebench-language17.2+0.2
apex-agents17.1+0.2
tau-bench17.1+0.2
arc-agi-316.8-0.2
swe-atlas-qna16.8-0.2
enigma-eval16.8-0.2
livebench-coding16.8-0.1
automationbench16.8-0.1
vibe-code16.8-0.1
itbench16.8-0.1
livebench-reasoning16.8-0.1
harvey17.0+0.1
aime16.9-0.1
swe-atlas-test-writing16.9-0.1
livecodebench16.9-0.1
vals-legal-research16.9-0.1
vals-code-migration16.9-0.1
vals-finance-agent16.9-0.0
vals-tax-agent16.9-0.0
math50017.0+0.0
gmmlu17.0+0.0
vals-excel-modeling16.9-0.0
swe-atlas-refactoring16.9-0.0
livebench-math16.9+0.0
frontiermath16.9+0.0
swe-bench-pro16.9+0.0
livebench-data16.9-0.0
What evidence is missing from the fit?

Across all collected settings for this model: 82 observations, 24 contributing, 0 matched without graph weight and 58 excluded. Counts do not establish rank eligibility.

  • 2 · Benchmark family or source protocol has not been reviewed
  • 1 · Original SciCode grading defects are under review; retained as supporting evidence, excluded from aggregate and capability scores. Corrected SciCode-Verified requires a separate protocol review. Sources: https://arxiv.org/abs/2608.04975 and https://artificialanalysis.ai/evaluations/scicode
  • 2 · The evaluator reports no named effort control. This does not identify a non-reasoning configuration.
  • 15 · Effort is not identified
  • 12 · No reviewed effort for this catalog observation; any separately collected effort measurement is counted separately
  • 25 · Supporting evidence outside the reviewed capability core
  • 1 · No matched opponent in this evaluation unit
Inspect every observation and exclusion →

Benchmark results

These are published benchmark scores. Supported effort estimates also appear on the Capability leaderboard. Results from different harnesses occupy different rows. Missing cells stay unknown; no scores are copied between effort levels.

Benchmark / sourceReasoningNon-reasoningUnspecified
AA-AnalystAgentArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Share of tasks solved on all 5 of 5 attempts · 80 tasks, 5 attempts per task · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

10%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
AA-Briefcase EloArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

AA-Briefcase v1.1 is an agentic knowledge work benchmark developed by Artificial Analysis. AA-Briefcase Elo is a combined metric that aggregates rubric pass rate, analytical quality Elo and presentation Elo · Higher is better · Evaluation results measured independently by Artificial Analysis

Unit: elo. Source collected 2026-10-18.

1,090.37 Elo
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
AA-LCR v1.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

83%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
AutomationBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Share of task objectives completed with no guardrail violations · Higher is better · Benchmark developed by Zapier · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

21.25%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
CritPtArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Benchmark developed by Argonne and UIUC, with contributions from 60+ researchers globally. CritPt is under review by its authors following external feedback · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

3.71%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
EnterpriseOps-Gym-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Overall task success rate across all eight enterprise domains (oracle tool mode) · Benchmark developed by ServiceNow Research · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

32.11%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
GDP.pdf: All-pass RateArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Share of attempts where every atomic criterion passed · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

9.8%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
GDPval-AA v2.1 LeaderboardArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Elo rating for performance on real-world work tasks · Anchored to DeepSeek V4.1 Flash (max) at 1600 · Higher is better · Evaluation results measured independently by Artificial Analysis

Unit: elo. Source collected 2026-10-18.

1,242.47 Elo
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
GPQA DiamondArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

92.93%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
Harvey LAB-AA v1.1: Hallucination-Gated All-Pass RateArtificial Analysis · Stirrup; Harvey private dataset v1.1.0; 120 tasks across 24 practice areas; up to 200 turns; internet disabled; GPT-6 Sol, Grok 4.7 and Claude Opus 5.5 rubric judges; GPT-6 Sol High two-pass hallucination check; hallucination-gated all-pass averaged across judges
Benchmark details

Average over tasks of the share of the judge panel finding every rubric criterion passed, with a task zeroed on any material hallucination · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

0%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
Humanity's Last ExamArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

38.97%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
IFBenchArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

82.86%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
MLCR-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Share of tasks judged accurate, complete, and concise · Expert + compound (hard) sets · Benchmark developed by Wisedocs · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

17.22%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
MMMU-ProArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

78.55%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
AA-Omniscience IndexArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. · Evaluation results measured independently by Artificial Analysis

Unit: index. Source collected 2026-10-18.

1.35
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
AA-Omniscience AccuracyArtificial Analysis · 6000 questions; 1 attempt; correct-answer share; GPT-5.6 Luna medium grader
Benchmark details

AA-Omniscience Accuracy (higher is better) measures the proportion of correctly answered questions out of all questions, regardless of whether the model chooses to answer · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

16.7%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
SciCodeArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Benchmark curated by scientists across 16 disciplines. SciCode is under review following an independent audit of the dataset · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

47.11%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
𝜏²-Bench TelecomArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

88.89%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
𝜏³-BankingArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Benchmark developed by Sierra Research · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

15.26%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
Terminal-Bench-Science 0.1Artificial Analysis · mini-swe-agent; Terminal-Bench-Science 0.1.0; 70 tasks; pass@1 averaged over 3 repeats
Benchmark details

Share of tasks passed (pass@1), averaged over three repeats · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

0.48%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
Terminal-Bench HardArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

42.42%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
Terminal-Bench v2.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Benchmark developed by the Laude Institute and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

65.17%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
Terminal-Bench v4.0Artificial Analysis · mini-SWE-agent v2.4.6; 66 tasks; pass@1 averaged over 3 repeats; 500 steps; task timeout up to 8 hours
Benchmark details

Benchmark developed by the Laude Institute, Stanford, and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-10-18.

2.02%
Reported settings

MiniMax-M3; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/minimax-m3; Mode from evaluator isReasoning metadata; effort budget not specified

Not reportedNot reported
FrontierCode 1.1 · extended-mini-swe-agentCognition · mini-swe-agent
Benchmark details

FrontierCode 1.1; mini-swe-agent. Open each result for its exact effort, task subset, metric and protocol details.

Unit: percent. Source collected 2026-10-18.

Not reportedNot reported
28.54%
Reported settings

FrontierCode 1.1; extended; 150 tasks; mini-swe-agent; source effort none; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0107

The evaluator reports no named effort control. This does not identify a non-reasoning configuration.

Source label needs review
FrontierCode 1.1 · main-mini-swe-agentCognition · mini-swe-agent
Benchmark details

FrontierCode 1.1; mini-swe-agent. Open each result for its exact effort, task subset, metric and protocol details.

Unit: percent. Source collected 2026-10-18.

Not reportedNot reported
14.7%
Reported settings

FrontierCode 1.1; main; 100 tasks; mini-swe-agent; source effort none; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.015

The evaluator reports no named effort control. This does not identify a non-reasoning configuration.

Source label needs review
What these comparisons mean

Each source reports its own effort labels. Matching labels do not establish equal compute budgets or identical fallback behavior. Open Reported settings to inspect the exact model name, agent and task coverage. Multiple results at one level remain visible rather than selecting the best. Source-reported None or Non-reasoning labels are not treated as evidence that an API supports those settings.

These measurements use their own dated source collection. The leaderboard fits identified configurations jointly; it contains no pooled model ratings. Explicit thinking budgets are separate configurations. Read the methodology.

Download all effort benchmark results · 8,916 results across 259 models · Collected 2026-10-18