Skip to content

MiniMax · Chat

MiniMax M3

MiniMax 🇨🇳ChatOne of its sources says the weights aren't publicly available.The maker lists this model as deprecated (6 Nov 2026).
From Empiriolabs's listing:
MiniMax M3 is a multimodal reasoning model for coding, agents, and long-context analysis with text, image, and video input.
llminfo IQ
126.8
Maker's price
$0.525 blended / 1M
Max context
1M
Listings
131

Evidence and sources

International API storefront (minimax.io) is Singapore-based; principal operating entity 上海稀宇科技有限公司 (Shanghai).

🤗 MiniMaxAI/Minimax-M3llminfo IQ 126.8, ranked among chat models. From 332 benchmarks. Plenty of data behind it.How llminfo IQ is computedCoding IQ 107.2, 95% interval 98–117. From 28 benchmark families across 14 sources. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.How Coding IQ is computedFair value (50.9). Costs per task about what its benchmark results predict. From 53 benchmarks with measured task costs. Plenty of data behind it.How Value for money worksYap score: tokens used per coding task vs. models with the same resultsChatty (71 ±8). Generates more tokens than models with similar coding results. Measured at default effort. A fair amount of data behind it. By effort: default 71 · none 54.Median 33,045 output tokens per task at default effort.How the Yap score worksReleased 31 May 2026Maker HQ: CN (source)
All rankings: #36 on OpenRouter this month #43 on OpenRouter today #48 on OpenRouter this week #97 on NanoGPT recentlySources: LMArena data, Ambient data, EmpirioLabs data, ModelRunner.ai data, LiteLLM data, llminfo researchPrimary data from LMArena from Ambient: name from EmpirioLabs: primary mode, description from ModelRunner.ai: weights availability from LiteLLM: deprecation date from llminfo research: release date
Tags from its sources:instructtoolsreasoningvisionlong-contextcodingagentic

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ

Trade-offs

Deprecated by its maker

The maker has already deprecated this model (6 Nov 2026).

No open weights

The weights aren't public, so you can't run it yourself.

We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 06:07 UTC.Prices come from the model's maker where they sell it, otherwise their official partner, otherwise the market median. This comparison uses 8 maker's prices; 2 market median prices.

Input & output

as stated by the model's providers
Text
Images
Video
Files
Text

Capabilities

Supported operations and provider-stated features

VisionReasoningFunction callingStructured outputPrompt cachingWeb search

Listings

 MiniMaxMiniMax-M3> 512k input tokens*Permanent 50% off · 512k-input-tokens-permanent-50-offSource: MiniMax API PricingNot published by the provider$1.20cache read $0.24$4.80
 Aihubmixcoding-minimax-m3-free · coding, freeSource: AIHubMix data docs1M$0.00$0.00
 Neosantaraminimax-m3-free · freeSource: Neosantara Inference data1M$0.00$0.00
 Vultrminimax-m3Source: Models.dev data524K$0.20$0.90
 Nous Portalminimax/minimax-m3:batch · batchSource: Nous Portal Inference Gateway data524K$0.224cache read $0.0448$0.88
 OpenRouterInference by DeepInframinimax/minimax-m3:batch · batch, fp8 · fp8Source: OpenRouter data docs524K$0.224cache read $0.0448$0.88
 Empiriolabsminimax-m3Sources: EmpirioLabs data, LiteLLM data1M$0.225cache read $0.045$0.90
 Hugging FaceInference by CoreweaveMiniMaxAI/MiniMax-M3Source: Hugging Face Inference Providers data262K$0.23$0.96
 Kilominimax/minimax-m3Sources: Kilo AI Gateway data, Models.dev data262K$0.23cache read $0.05$0.96
 Ambientminimax/minimax-m3Sources: Ambient data, Models.dev data131K$0.23cache read $0.05$0.96

11 listings without a price are hidden.

Included in plans

CheapestInference Frontier PoolcheapestinferenceSubscription$71/moPlan page
MiniMax M Plan GoMiniMaxSubscription$22/moPlan page
MiniMax M Plan ExploreMiniMaxSubscription$55/moPlan page
MiniMax M Plan BuildMiniMaxSubscription$132/moPlan page
OpenLimits MaxopenlimitsSubscription€120/moPlan page
OpenLimits UltimateopenlimitsSubscription€299/moPlan page

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Agentic

Arena AgentSource: LMArena data
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · MiniMax M2.7 grader
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure · rounded
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
BrowseComp3 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure · rounded
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
WebExplorer-style agent · history discarded beyond 64K
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Design Arena EloSource: Design Arena Leaderboards data · Nov 2026
Draco2 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
MiniMax Code Deep Research Skill · official rubrics · Opus 4.6 grader
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure · rounded
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
public GDPval cases · pointwise public rubrics
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
MCP Atlas3 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
public set · Gemini 2.5 Pro grader
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
OpenHandsSource: OpenHands Index data
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
nogdrive 361 · 1920×1080 · 200 steps
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure · setting not stated
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · Ralph-Loop · 12h
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
public dataset · Claude Code
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
SWE-Bench Pro4 related benchmarks · 7 results
SWE-Bench Pro
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · official-aligned evaluation
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
SWE Atlas - Codebase QnA
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
SWE Atlas - Test Writing
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · 4C8G · 3h · four-run mean
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
SWE-Atlas · Codebase Q&A
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Mini-SWE-Agent · 4C8G · 3h timeout
Source: OpenRouter benchmark exports
31 May 2026 version · 90-day results
Across 13 providers lowest 71.4% median 74.1% highest 76.1%OpenRouter's automatic routing 74.1%
Show results from 13 providers and OpenRouter's automatic routingHide results from 13 providers and OpenRouter's automatic routing
  • Together
  • SambaNova
  • Mara
  • CoreWeave
  • AtlasCloud
  • ModelRun
  • Minimax
  • OpenRouter's automatic routing
  • Parasail
  • Novita
  • GMICloud
  • Venice
  • StreamLake
  • DeepInfra
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Terminus 2 · 8C16G · 2h timeout · 128K max output
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
YC-Bench2 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official code/config · final assets
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Human preference

Arena ELO64 related benchmarks · 64 results
Source: LMArena data · Nov 2026
BenchmarkScore
Arena ELO
Chinese
Creative Writing
Document
Show 60 more benchmarksHide 60 benchmarks
BenchmarkScore
Document Style-Controlled
English
Exclude Ties
Expert
French
German
Hard Prompts
Hard Prompts English
Industry Business and Management and Financial Operations
Industry Entertainment and Sports and Media
Industry Life and Physical and Social Science
Industry Mathematical
Industry Medicine and Healthcare
Industry Software and It Services
Industry Writing and Literature and Language
Japanese
Korean
Longer Query
Multi Turn
Non English
Polish
Russian
Spanish
Style-Controlled
Text Factuality
Text Factuality - Chinese
Text Factuality - Creative Writing
Text Factuality - English
Text Factuality - Exclude Ties
Text Factuality - Expert
Text Factuality - French
Text Factuality - German
Text Factuality - Hard Prompts
Text Factuality - Hard Prompts English
Text Factuality - Industry Business and Management and Financial Operations
Text Factuality - Industry Entertainment and Sports and Media
Text Factuality - Industry Life and Physical and Social Science
Text Factuality - Industry Mathematical
Text Factuality - Industry Medicine and Healthcare
Text Factuality - Industry Software and It Services
Text Factuality - Industry Writing and Literature and Language
Text Factuality - Japanese
Text Factuality - Korean
Text Factuality - Longer Query
Text Factuality - Multi Turn
Text Factuality - Non English
Text Factuality - Polish
Text Factuality - Russian
Text Factuality - Spanish
Vision
Vision - Chinese
Vision - Creative Writing Vision
Vision - Diagram
Vision - English
Vision - Entity Recognition
Vision - Homework
Vision - Humor
Vision Style-Controlled
Design Arena EloSource: Design Arena Leaderboards data · Nov 2026

Coding

Arena ELOSource: LMArena data · Nov 2026
CL-benchSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
KernelBench3 related benchmarks · 5 results
Hard
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · NVIDIA Blackwell sm_120 · 9-task mean
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
LiveSQLBench2 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
LiveSQLBench-Base-Full v1 · 600 questions · Claude Code
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
NL2RepoSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
PaperBench2 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · Ralph-Loop · 12h · 19 papers · Opus 4.6 grader
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
SVGBench3 results
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
text/image inputs · VLM rendering verifier · three-run mean
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
SWE-fficiency6 related benchmarks · 8 results
SWE-fficiency
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · 1C2G · 2h
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
SWE-bench Verified
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · four-run mean
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
ValsSource: Vals AI
VIBE-V23 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · three-run mean
Reported by MiniMax · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official model-page overview figure · rounded
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

General

Science

GPQA Diamond16 results
Source: Epoch AI Benchmarking Hub data
Reasoning effort
Effort not reported
Source: OpenRouter benchmark exports
31 May 2026 version · 90-day results
Across 13 providers lowest 81.8% median 90.8% highest 91.6%OpenRouter's automatic routing 90.7%
Show results from 13 providers and OpenRouter's automatic routingHide results from 13 providers and OpenRouter's automatic routing
  • DeepInfra
  • Minimax
  • Together
  • StreamLake
  • Novita
  • GMICloud
  • AtlasCloud
  • CoreWeave
  • Venice
  • OpenRouter's automatic routing
  • ModelRun
  • Parasail
  • Mara
  • SambaNova

Multilingual

EKI EstonianSource: Keelemudelite mõõdupuu data docs · Aug 2026

Reasoning

OfficeQA Pro2 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Claude Code · filesystem delivery · exact match
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Vision

Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official react mode · environment description 256K
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
MMMU-Pro2 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official-aligned prompt
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Video-MME2 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
with subtitles · 1 FPS · ≤1024 frames
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
VideoMMMU2 results
Reported by MiniMax Research · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
1 FPS · ≤512 frames · judge scoring
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Math

IMO 2025Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim

Long context

Instruction following

OCR

Factuality

Knowledge

Speech

Other

BlenderWorld15 related benchmarks · 15 results
Art & Animation
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 30 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Architecture & Interiors
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 72 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Characters & Animals
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 25 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Fully satisfied rubric
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Image/video-input overall score
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 145 image or video reference tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Nature & Landscapes
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 21 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Show 9 more resultsHide 9 results
Overall score
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Procedural & Game Assets
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 10 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Products & Manufacturing
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 28 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Rendering & Simulation
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 17 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Science & Visualization
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 9 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Text-only overall score
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 78 text-only tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Vehicles & Machinery
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 11 category tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Visual-quality dimension score
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Weighted rubric score
Reported by BlenderWorld authors / Alibaba Token Hub · via BenchBoard docs
benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard
Nejumi 428 related benchmarks · 28 results
ALT_バイアス
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
ALT_倫理・道徳
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
ALT_制御性
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
ALT_堅牢性
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
ALT_毒性
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
ALT_真実性
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
Show 22 more resultsHide 22 results
AVG_jaster_0shot
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
AVG_jaster_2shots
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_アプリケーション開発
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_コーディング
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_一般的知識
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_基礎的言語性能
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_専門的知識
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_応用的言語性能
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_情報検索
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_意味解析
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_抽象的推論
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_推論能力
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_数学的推論
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_構文解析
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_知識・質問応答
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_翻訳
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_表現
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_論理的推論
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
GLP_関数呼び出し
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
TOTAL_SCORE
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
アラインメント(ALT)_AVG
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
汎用的言語性能(GLP)_AVG
Source: Nejumi Leaderboard 4 data docs
MiniMaxAI/MiniMax-M3: reasoning-enabled
PerceptionBench9 related benchmarks · 9 results
Overall accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
Attribute accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
Comparison accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
Contextual integration accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
Counting accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
Depth and 3D perception accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
Show 3 more resultsHide 3 results
Fine-grained recognition accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
Localization accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
Visual relation accuracy
Reported by Moonshot AI / PerceptionBench authors · via BenchBoard docs
benchmark-owner evaluation · 3,000 verified questions total · open-ended short answers · unified prompts · highest available reasoning budget; level not specified · main-table result; run count not specified · GPT-oss-120B judge
SkillsBench8 related benchmarks · 8 results
1.1 · With Skills
Reported by SkillsBench / BenchFlow · via BenchBoard docs
OpenHands · 87 tasks · 3 trials/task · max reasoning effort · with Skills
1.1 · Without Skills
Reported by SkillsBench / BenchFlow · via BenchBoard docs
OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills
Vals · With Skills · Accuracy
Reported by Vals AI · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Vals AI · updated 2026-09-27 · metadata version 1 · 87 tasks in trace charts · 3 trials/task/condition · per-task mean then task average · OpenHands-CLI@ · With Skills · system=minimax/MiniMax-M3 · provider=MiniMax · temperature=1 · top_p=0.95 · max_output_tokens=512000
Vals · With Skills · Mean cost / test
Reported by Vals AI · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Vals AI · updated 2026-09-27 · metadata version 1 · 87 tasks in trace charts · 3 trials/task/condition · per-task mean then task average · OpenHands-CLI@ · With Skills · system=minimax/MiniMax-M3 · provider=MiniMax · temperature=1 · top_p=0.95 · max_output_tokens=512000
Vals · With Skills · Mean duration / test
Reported by Vals AI · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Vals AI · updated 2026-09-27 · metadata version 1 · 87 tasks in trace charts · 3 trials/task/condition · per-task mean then task average · OpenHands-CLI@ · With Skills · system=minimax/MiniMax-M3 · provider=MiniMax · temperature=1 · top_p=0.95 · max_output_tokens=512000
Vals · Without Skills · Accuracy
Reported by Vals AI · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Vals AI · updated 2026-09-27 · metadata version 1 · 87 tasks in trace charts · 3 trials/task/condition · per-task mean then task average · OpenHands-CLI@ · Without Skills · system=minimax/MiniMax-M3 · provider=MiniMax · temperature=1 · top_p=0.95 · max_output_tokens=512000
Show 2 more resultsHide 2 results
Vals · Without Skills · Mean cost / test
Reported by Vals AI · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Vals AI · updated 2026-09-27 · metadata version 1 · 87 tasks in trace charts · 3 trials/task/condition · per-task mean then task average · OpenHands-CLI@ · Without Skills · system=minimax/MiniMax-M3 · provider=MiniMax · temperature=1 · top_p=0.95 · max_output_tokens=512000
Vals · Without Skills · Mean duration / test
Reported by Vals AI · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Vals AI · updated 2026-09-27 · metadata version 1 · 87 tasks in trace charts · 3 trials/task/condition · per-task mean then task average · OpenHands-CLI@ · Without Skills · system=minimax/MiniMax-M3 · provider=MiniMax · temperature=1 · top_p=0.95 · max_output_tokens=512000
SuperCLUESource: SuperCLUE data
Vals18 related benchmarks · 20 results
Show 14 more resultsHide 14 results
Immigration
Source: Vals AI
LegalBench
Source: Vals AI
MedCode
Source: Clinical Benchmarks data docs · Sep 2026
Setting not stated
Source: Vals AI
Setting not stated
MedScribe
Source: Clinical Benchmarks data docs · Sep 2026
Setting not stated
Source: Vals AI
Setting not stated

Data credits: SEA-HELM leaderboard, AI Singapore (AI Products team), National University of Singapore. Published aggregate scores; results licence not stated (SEA-HELM code is MIT). · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · LMArena, CC BY 4.0 (licence) · Data by LLM Stats · Baromètre de l’écriture IA, Johannes Eva; public numeric ranking data, © 2026 Johannes Eva, no results licence stated · BullshitBench by Peter Gostev, MIT (licence) · Hugging Face official benchmark leaderboards; licence varies by benchmark · CyberGym leaderboard (Wang, Shi, He, Cai, Zhang, Song; UC Berkeley), ICLR 2026. Results evaluated and submitted by the listed teams. No licence is stated for the result data; the repository's Apache-2.0 licence covers its code only. · Deep20Bench edition 1.1, Patrick Heusser. Results CC BY 4.0; native question scores and fraction-to-percent success rates. (licence) · Keelemudelite mõõdupuu: Independent LLM leaderboard for Estonian. Eesti Keele Instituut. https://moodupuu.eki.ee, CC BY 4.0. Latest per-benchmark results selected with partial-run flags retained. (licence) · EQ-Bench (Samuel J. Paech) — cite paech2023eqbench · OpenRouter public benchmark leaderboard and search-lane exports, CC BY 4.0. Fractional source scores converted to percentages; immutable dated snapshots. (licence) · Nejumi Leaderboard 4, Weights & Biases Japan; public publisher-selected numeric evaluation results · OpenHands Index (All Hands AI), Apache-2.0 (licence) · Steel.dev Agent Leaderboards, MIT (licence) · SnakeBench (Greg Kamradt / ARC Prize Foundation), MIT (licence) · Clinical Benchmarks (clinicalbenchmarks.ai), CC BY 4.0; fractions converted to percentages; source documents retain their own terms (licence)

Also known as: accounts/fireworks/models/minimax-m3, aihubmix/minimax-m3, azure_ai/fw-minimax-m3, cc-minimax-m3, cline-pass/minimax-m3, coding-minimax-m3, coding-minimax-m3-free, deepinfra/minimax-m3, deepinfra/minimaxai/minimax-m3, empiriolabs/minimax-m3, empiriolabs/minimax-m3:priority, fireworks/minimax-m3, fireworks_ai/accounts/fireworks/models/minimax-m3, fireworks_ai/minimax-m3, gmi/minimaxai/minimax-m3, minimax m3 preview, minimax-m3-512k-input-tokens-permanent-50-off, minimax-m3-512k-input-tokenspermanent-50-off, minimax-m3-free, minimax-m3-preview, minimax-m3:fast, minimax-m3@eu, minimax/minimax-m3, minimax/minimax-m3:batch, minimax/minimax-m3:thinking, minimaxai/minimax-m3, minimaxai/minimax-m3-mxfp8, minimaxi/minimax-m3, modelrunner-ai/minimax-m3, nebius/minimaxai/minimax-m3, novita/minimax-m3, novita/minimax/minimax-m3, opencode_go/minimax-m3, parasail-minimax-m3, parasail-minimax-m3-fast, streamlake/minimax-m3, tencent/minimax-m3, tensorx/minimax-m3, together_ai/minimaxai/minimax-m3, wandb/minimaxai/minimax-m3

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works