Skip to contentT/TTokens or Towers

Compare

LeaderboardEvery tracked model, ranked and priced side by sideBenchmarksWhat each test measures and who leads itPricesAPI, media, and hardware rates with a cost calculator

Plan

Stack plannerSix questions to a priced local and hosted stackShortlistOur picks for subscriptions, APIs, and hardwareEnterpriseSeat plans against self-hosting for a whole team

Build

HardwareGPUs and workstations, and which models they fitHostsWhere to rent open-weight inference, and for how muchHarnessesCoding agents and the models they drive

Lab

FindingsResults from our own evaluationsScanAudit a repo’s AI usage in the browserMethodWhere the numbers come from and how we check them
Data checked September 25, 2026
Data checked September 25, 2026
Plan a stack
T/TTokens or Towers

An independent guide to AI models, what they cost, and the hardware to run them. No account, no API key, no sponsors.

Compare

  • Leaderboard
  • Benchmarks
  • Prices

Plan

  • Stack planner
  • Shortlist
  • Enterprise

Build

  • Hardware
  • Hosts
  • Harnesses

Lab

  • Findings
  • Scan
  • Method
Data checked September 25, 2026 · Every number links to its sourceRSS · Source on GitHub
Leaderboard/MiniMax
MiniMax

MiniMax · MiniMax M3

MiniMax-M3

Open weightsGenerally availablecustom (MiniMax Community License)

MiniMax's current flagship LLM: an open-weight 428B/23B MoE with MiniMax Sparse Attention, 1M context and native image/video input, at a low per-token price.

AA Intelligence
29#41 of 142
Input / output per 1M
$0.30/ $1.20
Output speed
179tok/s
Context window
1M tokens

Composite indexes

  • Artificial Analysis Intelligence Index#41 of 142
    29independentv4.3.2, reasoningArtificial Analysis: MiniMax-M3 ↗

    Leader: Claude Opus 5.5 at 58

Coding

  • Terminal-Bench 4.0#40 of 88
    2%independentreasoning, AA-runArtificial Analysis: MiniMax-M3 ↗

    Leader: Claude Mythos 5.1 at 60.9%

  • Terminal-Bench 2.x#41 of 103
    66%vendorTerminal-Bench 2.1MiniMax blog: MiniMax M3 ↗
    65.2%independentTerminal-Bench 2.1, reasoning, AA-runArtificial Analysis: MiniMax-M3 ↗

    Leader: Claude Fable 5.1 at 91.4%

  • SciCode#50 of 94
    47.1%independentreasoning, AA-runArtificial Analysis: MiniMax-M3 ↗

    Leader: Claude Opus 5.5 at 66.9%

  • SWE-bench Pro#14 of 31
    59%vendorMiniMax blog: MiniMax M3 ↗

    Leader: Claude Opus 5.5 at 89.9%

Agents and tool use

  • τ²-Bench#22 of 88
    88.9%independenttelecom, reasoning, AA-runArtificial Analysis: MiniMax-M3 ↗

    Leader: GLM-5.2 at 99.1%

  • GDPval-AA#37 of 103
    1230independentGDPval-AA v2.1, reasoningArtificial Analysis: MiniMax-M3 ↗

    Leader: Claude Opus 5.5 at 1846

  • MCP Atlas#7 of 9
    74.2%vendorMiniMax blog: MiniMax M3 ↗

    Leader: Kimi K3 at 84.2%

  • OSWorld-Verified#11 of 16
    70.1%vendorOSWorld-VerifiedMiniMax blog: MiniMax M3 ↗

    Leader: Kimi K3 at 84.8%

Reasoning and knowledge

  • GPQA Diamond#16 of 142
    92.9%independentreasoning, AA-runArtificial Analysis: MiniMax-M3 ↗

    Leader: GPT-6 Astra at 96.1%

  • Humanity’s Last Exam#38 of 147
    39%independentreasoning, no tools, AA-runArtificial Analysis: MiniMax-M3 ↗

    Leader: Claude Opus 5.5 at 61.4%

  • CritPt#54 of 139
    3.7%independentreasoning, AA-runArtificial Analysis: MiniMax-M3 ↗

    Leader: GPT-5.6 Sol at 32.3%

Vision and long context

  • AA Long Context Reasoning#15 of 139
    83%independentreasoning, AA-LCRArtificial Analysis: MiniMax-M3 ↗

    Leader: Kimi K3 at 88.7%

  • MMMU-Pro#24 of 72
    78.6%independentreasoning, AA-runArtificial Analysis: MiniMax-M3 ↗

    Leader: Claude Opus 5.5 at 87.7%

Human preference

  • LMArena Text#38 of 54
    1441independentarena entry "minimax-m3", 48,540 votes, checked 2026-09-25Arena (LMArena) text leaderboard ↗

    Leader: Claude Fable 5 at 1506

  • LMArena WebDev#39 of 54
    1485independentarena entry "minimax-m3", 14,971 votes, checked 2026-09-25Arena (LMArena) WebDev/Code leaderboard ↗

    Leader: Claude Opus 5.5 at 1818

Other published results

ResultScoreReported bySource
IFBench (AA) · reasoning82.9%independentArtificial Analysis: MiniMax-M3 ↗
Terminal-Bench Hard (AA) · reasoning42.4%independentArtificial Analysis: MiniMax-M3 ↗
SWE-fficiency34.8%vendorMiniMax blog: MiniMax M3 ↗
KernelBench Hard28.8%vendorMiniMax blog: MiniMax M3 ↗

Specs

Released
June 1, 2026
Parameters
428B total, 23B active
Size class
100 to 500B
Input
text, image, video
Output
text
Context
1M tokens
Max output
Not published
Cached input
$0.06 / 1M
Blended 70/30
$0.57 / 1M
First token
1.07s
  • Price: MiniMax pay-as-you-go pricing ↗ Standard tier, 512K input or less, after a "permanent 50% off" (list $0.60 / $2.40). Above 512K input: $0.60 / $2.40. The Priority tier costs 1.5x.
  • Speed: Artificial Analysis: MiniMax-M3 ↗

Will it fit your budget?

The planner prices this model against local hardware and other APIs for your workload.

Plan a stack →

Compare with

  • AlibabaQwen3.7-Max29
  • GoogleGemini 3.1 Pro Preview29.7
  • Z.aiGLM-528
  • Thinking Machines LabInkling-Small27.8