← All models

MiniMax-M2.7

open weightsdeprecated by benchmark source
MiniMax · released 2026-03-18 · 1 offer

Context 205K tokens

Top 1 cheapest providers (Adjusted $/task)

Within the active global provider, residency and confidentiality filters.

#ProviderAdjusted $/task
1Groqvia OpenRouter — Price via OpenRouter: OpenRouter's price for Groq's endpoint (https://openrouter.ai/minimax/minimax-m2.7/providers), collected 2026-11-21.

Composite

2 of 7 inputs6 radar axes: DesignArena's two boards share one

20.8

 Thin data · 2/7Based on only 2 of 7 Composite inputs — treat this rank as uncertain

includes −10.0 for its Benchmaxxing signal (from 30.8; why, switch off in Options)

Benchmaxxing signal +10.0, medium Benchmaxxing tag · medium report →

AA Coding v4.xCoding Agent v1.4AA Intelligence (est.)AA AgenticEpoch ECISoftware ECIDesignArenaAA Coding v4.x: percentile 60AA Intelligence (est.): percentile 70AA Agentic: percentile 52

AA Coding v4.x 52.6Coding Agent v1.4 —AA Intelligence (est.) 11.8AA Agentic 16.8Epoch ECI —Software ECI —DesignArena —/—

Radar: percentile among all models measured on each input; a gap means not measured.

Benchmark sheet

41 of 320 registered benchmark versions · bars show the percentile among all models measured on each benchmark.

Compare this model →

Agentic

  • AA-AnalystAgent1911.3%

    AA-AnalystAgent published 2026-09-10 → · Published board — Tests analyst tasks using agentic Python execution across fourteen domains.

  • APEX-Agents-AA1410.6%

    APEX-Agents-AA published 2026-09-10 → · Published board — Tests professional-service tasks that require agents to produce locally graded deliverables.

  • AA-Briefcase33715 Elo

    AA-Briefcase published 2026-09-10 → · Published board — Tests multi-week professional knowledge-work projects with linked tasks and large source collections.

  • GDPval-AA v2551,087 Elo

    GDPval-AA v2 v2 → · Published board — Tests professional knowledge-work deliverables across occupations using AA's Stirrup harness.

  • ITBench-AA2826.5%

    ITBench-AA published 2026-09-10 → · Published board — Tests root-cause diagnosis from offline Kubernetes incident snapshots.

  • Terminal-Bench Hard (AA)8739.4%

    Terminal-Bench Hard (AA) pinned revision 74221fb → · Published board — Tests a pinned 44-task hard subset of terminal-based work using Terminus 2.

  • Terminal-Bench v2.1 (AA)5555.4%

    Terminal-Bench v2.1 (AA) v2.1 → · Published board — Tests terminal-based work on the 89-task verified refresh using Terminus 2.

  • Terminal-Bench v4.0 (AA)180.0%

    Terminal-Bench v4.0 (AA) v4.0 → · Published board — Tests terminal-based work on the 66-task release using mini-SWE-agent v2.4.6.

  • τ²-Bench Airline (OpenRouter run)5371.1%

    τ²-Bench Airline (OpenRouter run) published 2026-09-15 → · Published board — Multi-turn service agents making tool calls under strict policy constraints. (OpenRouter's own reproducible run, as published on openrouter.ai/benchmarks.)

  • Finance Agent v2 (Vals Index v2)027.9%

    Finance Agent v2 (Vals Index v2) v2 → · Published board — Multi-step financial reasoning tasks.

  • HLAB — Harvey's Legal Agent Benchmark (Vals Index v2)00.0%

    HLAB — Harvey's Legal Agent Benchmark (Vals Index v2) v2 → · Published board — Long-horizon legal work product creation. The index column is Harvey's Legal Agent Benchmark's own published standalone score, reported as accuracy — not comparable with Artificial Analysis' Harvey LAB-AA row.

  • Terminal-Bench 2.1 (Vals Index v2)348.7%

    Terminal-Bench 2.1 (Vals Index v2) v2 → · Published board — Command-line interface problem solving; the page states that each index column is that benchmark's own published standalone score, so the runner is not asserted here.

  • Vals Index v2 (Vals AI)324.6%

    Vals Index v2 (Vals AI) v2 → · Published board — GDP-weighted average of agentic model performance across finance, coding and legal tasks.

  • AA Agentic Index5216.8

    AA Agentic Index published 2026-09-18 → · Published board — Artificial Analysis publishes one Agentic Index per model, measured on its primary configuration and relayed by OpenRouter's Benchmarks API; the value is attached at family scope on the deterministic representative. Artificial Analysis publishes this board without a version number; we keep the date each result was retained.

Coding

  • SciCode (AA subproblems) v1.0.15450.1%

    SciCode (AA subproblems) v1.0.1 → · Published board — Tests scientific Python programming with scientist-annotated background information.

  • Code Migration (Vals Index v2 subset)310.0%

    Code Migration (Vals Index v2 subset) v2 → · Published board — Porting projects to another language, including COBOL modernization.

  • Vibe Code Bench (Vals Index v2)011.9%

    Vibe Code Bench (Vals Index v2) v2 → · Published board — End-to-end app-building tasks.

  • WeirdML v23336.9%

    WeirdML v2 → · Published board — Presents LLMs with unusual machine-learning tasks where they must write, run, and iteratively improve PyTorch code under fixed compute and time limits.

  • AA Coding Index v4.x · frozen 2026-10-276052.6

    AA Coding Index v4.x · frozen 2026-10-27 → · Published board — Artificial Analysis withdrew the Coding Index with Intelligence Index v5.0 (2026-10-28). These are the last v4.x values, frozen with their own date (2026-10-27).

Efficiency

Instruction-following

Knowledge

Long-context

  • GDP.pdf (AA)356.0%

    GDP.pdf (AA) published 2026-09-10 → · Published board — Tests professional reasoning over long PDFs with AA document preparation and grading.

  • AA-LCR v1.18278.3%

    AA-LCR v1.1 v1.1 → · Published board — Tests reasoning across multiple long documents with corrected answer keys and grading.

  • MLCR-AA143.3%

    MLCR-AA published 2026-09-10 → · Published board — Tests medical-record synthesis and reasoning across long, fragmented case documents.

  • Context Arena — MRCR v2 (8-needle)934.1%

    Context Arena — MRCR v2 (8-needle) 8-needle → · Published board — The model must find and reproduce the right one of eight near-identical earlier replies hidden in a long conversation, averaged over context lengths from 8k to 128k tokens.

Science

Tool-use

  • AutomationBench-AA v1.0.6284.4%

    AutomationBench-AA v1.0.6 → · Published board — Tests multi-app SaaS workflows through REST tools on a held-out AutomationBench split.

  • τ²-Bench Telecom (AA)7784.8%

    τ²-Bench Telecom (AA) published 2026-09-10 → · Published board — Tests dual-control telecom agents that coordinate tool use with a simulated user.

  • τ³-Banking (AA) v1.0.1309.9%

    τ³-Banking (AA) v1.0.1 → · Published board — Tests banking support agents that retrieve policies and change account state through tools.

  • Toolathlon-Verified (HKUST NLP) 2026-06-30647.5%

    Toolathlon-Verified (HKUST NLP) 2026-06-30 → · Published board — An agent completes 108 real-world tool-use tasks across 32 MCP servers and 7 local toolkits, and each task is graded by executing the resulting state against the task's ground truth.

  • Excel Modeling Benchmark (Vals Index v2)328.5%

    Excel Modeling Benchmark (Vals Index v2) v2 → · Published board — Building and editing financial models in spreadsheets.

Uncensored

  • UGI Willingness31.00

    UGI Willingness published 2026-09-10 → · Published board — A private-question evaluation of how far a model follows challenging instructions before refusing or deviating.

  • UGI1318.0

    UGI published 2026-09-10 → · Published board — A private-question suite combining sensitive-topic knowledge with willingness to follow controversial instructions.

Writing

  • LLM Creative Story-Writing Benchmark0-4.03

    LLM Creative Story-Writing Benchmark published 2026-09-10 → · Published board — Pairwise comparison of short stories written to the same constrained creative briefs, with LLM evaluator choices combined into a relative comparison score.

  • UGI Writing5537.6

    UGI Writing published 2026-09-10 → · Published board — A maintainer-owned writing evaluation considering intelligence, style, repetition and length adherence, informed by human preferences.

Unusual results

6 threshold-crossing signals flagged · 39 eligible benchmark families

How flags are calculated

Heuristic screen, not statistical significance: benchmark families are correlated and source uncertainty is unknown. Peer evidence needs ≥ 20 independently measured matched configurations from ≥ 10 distinct model families. A flag needs a directed population z-score of magnitude ≥ 1.5 and a gap of ≥ 1.5 from the leave-one-benchmark-family-out mean z in the same direction, over ≥ 5 other benchmark families.

  • unusually weak Finance Agent v2 (Vals Index v2)
    Why
    Observed: 27.887 percent
    Peer mean 52.22568 · peer sd 7.31503 · n 38 · families 37
    Directed z -3.327 · baseline z -0.391 · gap -2.936 · profile n 38
    27.887 percentmeasuredobserved 2026-09-26vals.ai ↗
    Evidence
    Axis: Finance Agent v2 (Vals Index v2) · v2 · Published board
    Exact value: 27.887 percent
    Observed: 2026-09-26T04:22:13.903077+00:00 · publication date: not recorded
    Observation id: public:098ca5ca32f3cf1a595cc022
  • unusually weak Vibe Code Bench (Vals Index v2)
    Why
    Observed: 11.926 percent
    Peer mean 68.67 · peer sd 20.89215 · n 39 · families 38
    Directed z -2.716 · baseline z -0.407 · gap -2.309 · profile n 38
    11.926 percentmeasuredobserved 2026-09-26vals.ai ↗
    Evidence
    Axis: Vibe Code Bench (Vals Index v2) · v2 · Published board
    Exact value: 11.926 percent
    Observed: 2026-09-26T04:22:13.903077+00:00 · publication date: not recorded
    Observation id: public:0f5b3e026a9744e5a089a8e4
  • unusually weak Vals Index v2 (Vals AI)
    Why
    Observed: 24.556 percent
    Peer mean 54.71374 · peer sd 11.34419 · n 38 · families 37
    Directed z -2.658 · baseline z -0.408 · gap -2.25 · profile n 38
    24.556 percentmeasured· 0.88 USD per rolloutobserved 2026-09-26vals.ai ↗
    Evidence
    Axis: Vals Index v2 (Vals AI) · v2 · Published board
    Exact value: 24.556 percent
    Published mean cost: 0.87768 USD per rollout
    Observed: 2026-09-26T04:22:13.903077+00:00 · publication date: not recorded
    Observation id: public:f24e8ba4a557f3c95b4b2a97
  • unusually strong IFBench (AA single-turn) published 2026-09-10
    Why
    Observed: 0.75714 fraction
    Peer mean 0.48006 · peer sd 0.17139 · n 450 · families 331
    Directed z 1.617 · baseline z -0.521 · gap 2.138 · profile n 38
    0.75714 fractionmeasuredobserved 2026-09-10artificialanalysis.ai ↗
    Evidence
    Axis: IFBench (AA single-turn) · published 2026-09-10 · Published board
    Exact value: 0.757142857142857 fraction
    Observed: 2026-09-10T21:47:16.627Z · publication date: not recorded
    Observation id: aa:4bbceacb-cf47-464b-b60f-e1d1fe016d67:ifbench
  • unusually weak Excel Modeling Benchmark (Vals Index v2)
    Why
    Observed: 28.451 percent
    Peer mean 60.6871 · peer sd 13.03063 · n 40 · families 39
    Directed z -2.474 · baseline z -0.413 · gap -2.061 · profile n 38
    28.451 percentmeasuredobserved 2026-09-26vals.ai ↗
    Evidence
    Axis: Excel Modeling Benchmark (Vals Index v2) · v2 · Published board
    Exact value: 28.451 percent
    Observed: 2026-09-26T04:22:13.903077+00:00 · publication date: not recorded
    Observation id: public:e30f2faa14bfd1250f51ec09
  • unusually weak Legal Research Bench (Vals Index v2)
    Why
    Observed: 10.577 percent
    Peer mean 37.35572 · peer sd 11.54643 · n 40 · families 39
    Directed z -2.319 · baseline z -0.417 · gap -1.902 · profile n 38
    10.577 percentmeasuredobserved 2026-09-26vals.ai ↗
    Evidence
    Axis: Legal Research Bench (Vals Index v2) · v2 · Published board
    Exact value: 10.577 percent
    Observed: 2026-09-26T04:22:13.903077+00:00 · publication date: not recorded
    Observation id: public:7f1e982cf5224b98dcd7b0da
Missing coverage · 279 benchmark versions

No result does not mean a zero, or that the model was never tested. Collection failures and disputed versions retain their distinct status.

Token offers by platform · 1 offer (Adjusted $/task)

Click any underlined price to see how it is estimated and where each input comes from. How we calculate adjusted cost.

1 offer within the active global filters; “price n/a” means the provider serves the model but publishes no direct token price we could collect — no other channel's price is shown in its place.

OpenRouter (1)

Prices via OpenRouter — what OpenRouter charges for each provider's endpoint, not the provider's own direct price. Source: openrouter.ai/minimax/minimax-m2.7/providers, collected 2026-11-21.

Groq
MiniMax-M2.7 — benchmarks & cost | Benchmark Heaven