NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/MiniMax M3
minimax logo

MiniMax M3

by minimax · Released May 2026

Open SourceMultimodal1M Context
55.1
avg score
Rank #118
Compare
Better than 62% of all models
Context
1.0M tokens (~524 books)
Input $/1M
$0.30
Output $/1M
$1.20
Type
multimodal
License
Open Source
Benchmarks
35 tested
Data as of
About

MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding,...

Tested on 35 benchmarks · BenchGecko score 55.1. Top scores: Chatbot Arena Elo — Coding (1482.0%), Chatbot Arena Elo — Overall (1440.1%), Artificial Analysis · GPQA Diamond (92.9%).

Looking for similar performance at lower cost?
Qwen3 235B A22B Thinking 2507 scores 55.3 (100% as good) at $0.23/1M input · 23% cheaper
Capabilities
coding
47.6
#101 globally
reasoning
61.9
#55 globally
math
74.0
#35 globally
knowledge
55.8
#87 globally
agentic
37.7
#32 globally
language
67.2
#88 globally
general
30.4
#123 globally
speed
55.3
#18 globally
Benchmark Scores
Compare All
Tested on 35 benchmarks · Ranked across 9 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
codingCompare coding →
LiveBench — Coding

Regularly refreshed coding problems that avoid data contamination. New problems added monthly to prevent memorization.

68.2·
LiveBench — Agentic Coding

LiveBench coding tasks that require multi-step reasoning and tool use. Tests planning and execution of complex coding workflows.

60.0·
Frontiercode
14.7·
reasoningCompare reasoning →
LiveBench — Data Analysis

Fresh data analysis tasks testing ability to interpret tables, charts, and statistical data.

76.2·
LiveBench — Reasoning

Regularly refreshed reasoning problems testing logical deduction, spatial reasoning, and analytical thinking.

74.5·
SimpleBench

Deceptively simple questions that humans find easy but AI models often get wrong. Tests common sense and reasoning gaps.

35.0·
mathCompare math →
LiveBench — Mathematics

Regularly updated math problems that test numerical reasoning, algebra, calculus, and combinatorics.

77.0·
OTIS Mock AIME 2024-2025

Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.

71.1·
Quick compare:DeepSeek logovs DeepSeek V3OpenAI logovs o3Alibaba Qwen logovs Qwen3 235B A22B
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · minimax MiniMax M3
May 2026
55.1
$0.30/M in1.0Mctx35 benchmarks
May 2026
$0.30/M in524Kctx(-524K)
May 2026
$0.00/M in(-0.30)1.0Mctx(+524K)
See the full MiniMax M3 family →
Similar Models
DeepSeek logo
DeepSeek V3
DeepSeek
55.1$0.26/1M
OpenAI logo
o3
OpenAI
55.1$2.00/1M
Alibaba Qwen logo
Qwen3 235B A22B
Alibaba Qwen
55.2$0.46/1M
Links
Info
minimaxPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@minimax
Source Code
GitHubHugging Face
BenchGecko API
minimax-m3
Specifications
  • Typemultimodal
  • Context1.0M tokens (~524 books)
  • ReleasedMay 2026
  • LicenseOpen Source
  • StatusActive
  • Cost / Message~$0.002
Available On
minimax logominimax$0.30
Categories
codingreasoningmathknowledgeagenticlanguagegeneralspeed
Learn More
context-windowtransformeropen weightstokens
Share & Export
Tweet
Related Models
DeepSeek logoDeepSeek V3
55.1
OpenAI logoo3
55.1
Alibaba Qwen logoQwen3 235B A22B
55.2
OpenAI logoGPT-5.1
55.0
MiniMax M3 is an open-source multimodal AI model by minimax, released in May 2026. It has an average benchmark score of 55.1. Context window: 1M tokens.

Related Models

DeepSeek V3 · DeepSeeko3 · OpenAIQwen3 235B A22B · Alibaba QwenGPT-5.1 · OpenAIQwen3 235B A22B Thinking 2507 · Alibaba Qwen

Benchmarks

Chatbot Arena Elo — CodingChatbot Arena Elo — OverallArtificial Analysis · GPQA DiamondArtificial Analysis · tau2-Bench TelecomGPQA diamond

Related Pages

minimax · Providerminimax · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-10-05

  • MiniMax M3 by minimax. BenchGecko score 55.1, rank 116 of 312 scored models (normalized average of public benchmark scores).
  • List price $0.30 input · $1.20 output per 1M tokens (as of 2026-10-05).
  • Gecko Tests · Gecko Score 67 (rank 2): Who Are You B (Knows who made it) · World Map C (92.8% of the map right) · Tokenizer Tax A (36% more tokens outside English).

How to cite · data as of 2026-10-05

MiniMax M3 · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-05. https://benchgecko.ai/model/minimax-m3

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP