# MiniMax M3 As of October 22, 2026, MiniMax M3 ranks #66 of 123 models overall with 51.7 ±13.0 on the LLMBenchmark Index. MiniMax M3 ranks #35 of 122 in reasoning, its strongest category. MiniMax M3 costs $0.30 per million input tokens and $1.20 per million output tokens. MiniMax M3 was released on May 31, 2026. Page: https://llmbenchmark.ai/models/minimax-minimax-m3 Data from October 22, 2026. Every figure below names the board it was read from and the day. - Maker: MiniMax - Released: May 31, 2026 - Price per million input tokens: $0.30 - Price per million output tokens: $1.20 - Context window: 1.05M tokens - Overall rank: 66 of 123, by the LLMBenchmark Index - LLMBenchmark Index: 51.7 ±13.0 (90% margin, from 7 of 7 inputs) - Artificial Analysis Intelligence Index: 29.2 - Vibe Coding Index: 39.2 (from 6 of 6 coding benchmarks) Benchmark results (116 of 195): - LiveBench coding: 70.9% (source: LiveBench, read October 22, 2026) - LMArena WebDev: 1482 Elo (source: LMArena WebDev, read October 22, 2026) - GPQA Diamond: 92.7% (source: GPQA (Vals AI), read October 22, 2026) - MMLU Pro: 84.2% (source: MMLU Pro (Vals AI), read October 22, 2026) - AA Intelligence Index: 29.2 (source: Artificial Analysis (via OpenRouter), read October 22, 2026) - AA Coding Index: 58.6 (source: Artificial Analysis (via OpenRouter), read October 22, 2026) - AA Agentic Index: 29.5 (source: Artificial Analysis (via OpenRouter), read October 22, 2026) - Design Arena: websites: 1257 Elo (source: Design Arena, read October 22, 2026) - Design Arena: UI components: 1242 Elo (source: Design Arena, read October 22, 2026) - Design Arena: game dev: 1210 Elo (source: Design Arena, read October 22, 2026) - Design Arena: data viz: 1236 Elo (source: Design Arena, read October 22, 2026) - Design Arena: 3D: 1208 Elo (source: Design Arena, read October 22, 2026) - Design Arena: all code categories: 1247 Elo (source: Design Arena, read October 22, 2026) - Design Arena: SVG: 1161 Elo (source: Design Arena, read October 22, 2026) - Design Arena: ASCII art: 1161 Elo (source: Design Arena, read October 22, 2026) - Design Arena: agents building web apps: 1200 Elo (source: Design Arena, read October 22, 2026) - Design Arena: agents building full-stack apps: 1197 Elo (source: Design Arena, read October 22, 2026) - Design Arena: agents building mobile apps: 1169 Elo (source: Design Arena, read October 22, 2026) - Design Arena: agents building Android apps: 1159 Elo (source: Design Arena, read October 22, 2026) - Design Arena: agents building games: 1158 Elo (source: Design Arena, read October 22, 2026) - Design Arena: HTML slides: 1130 Elo (source: Design Arena, read October 22, 2026) - Design Arena: PowerPoint slides: 1181 Elo (source: Design Arena, read October 22, 2026) - Design Arena: spreadsheets: 1028 Elo (source: Design Arena, read October 22, 2026) - LiveBench reasoning: 74.5% (source: LiveBench, read October 22, 2026) - LiveBench data analysis: 76.2% (source: LiveBench, read October 22, 2026) - LiveBench research: 68.0% (source: LiveBench, read October 22, 2026) - LiveBench browser: 8.0% (source: LiveBench, read October 22, 2026) - LiveBench long agentic: 12.4% (source: LiveBench, read October 22, 2026) - LiveBench design: 66.2% (source: LiveBench, read October 22, 2026) - LiveCodeBench (Vals AI): 82.2% (source: Vals AI boards, read October 22, 2026) - SWE-bench Verified (Vals AI): 75.0% (source: Vals AI boards, read October 22, 2026) - Terminal-Bench 4 (Vals AI): 1.0% (source: Vals AI boards, read October 22, 2026) - Vibe Code Bench (Vals AI): 47.6% (source: Vals AI boards, read October 22, 2026) - Code migration (Vals AI): 19.9% (source: Vals AI boards, read October 22, 2026) - Terminal-Bench 2.1 (Vals AI): 53.6% (source: Vals AI boards, read October 22, 2026) - SWE-bench Multimodal (Vals AI): 33.3% (source: Vals AI boards, read October 22, 2026) - Vibe Code Bench 1-100 (Vals AI): 9.2% (source: Vals AI boards, read October 22, 2026) - SkillsBench (Vals AI): 51.5% (source: Vals AI boards, read October 22, 2026) - CyberBench (Vals AI): 65.0% (source: Vals AI boards, read October 22, 2026) - Finance Agent (Vals AI): 48.3% (source: Vals AI boards, read October 22, 2026) - Tax Agent Bench (Vals AI): 20.6% (source: Vals AI boards, read October 22, 2026) - Legal Research Bench (Vals AI): 29.8% (source: Vals AI boards, read October 22, 2026) - Excel Modeling Benchmark (Vals AI): 47.8% (source: Vals AI boards, read October 22, 2026) - Public Benefits Bench (Vals AI): 64.1% (source: Vals AI boards, read October 22, 2026) - MMMU Pro (Vals AI): 81.2% (source: Vals AI boards, read October 22, 2026) - SAGE (Vals AI): 50.6% (source: Vals AI boards, read October 22, 2026) - LegalBench (Vals AI): 85.4% (source: Vals AI boards, read October 22, 2026) - CorpFin (Vals AI): 68.1% (source: Vals AI boards, read October 22, 2026) - TaxEval (Vals AI): 72.7% (source: Vals AI boards, read October 22, 2026) - MortgageTax (Vals AI): 68.4% (source: Vals AI boards, read October 22, 2026) - MedCode (Vals AI): 46.3% (source: Vals AI boards, read October 22, 2026) - MedScribe (Vals AI): 87.3% (source: Vals AI boards, read October 22, 2026) - Vals Multimodal Index: 60.0% (source: Vals AI boards, read October 22, 2026) - GPQA Diamond (Artificial Analysis run): 92.9% (source: Artificial Analysis, read October 22, 2026) - Humanity's Last Exam (Artificial Analysis run): 39.0% (source: Artificial Analysis, read October 22, 2026) - SciCode: 47.1% (source: Artificial Analysis, read October 22, 2026) - Terminal-Bench Hard (Artificial Analysis run): 42.4% (source: Artificial Analysis, read October 22, 2026) - IFBench: 82.9% (source: Artificial Analysis, read October 22, 2026) - Long context reasoning (AA-LCR): 83.0% (source: Artificial Analysis, read October 22, 2026) - τ²-Bench Telecom: 88.9% (source: Artificial Analysis, read October 22, 2026) - FrontierCode: 14.7% (source: Epoch AI benchmarking hub, read October 22, 2026) - APEX-Agents: 37.7% (source: Epoch AI benchmarking hub, read October 22, 2026) - OSWorld 2.0: 4.6% (source: Epoch AI benchmarking hub, read October 22, 2026) - GPQA Diamond (Epoch AI run): 90.9% (source: Epoch AI benchmarking hub, read October 22, 2026) - SimpleBench: 45.8% (source: Epoch AI benchmarking hub, read October 22, 2026) - Chess puzzles: 13.0% (source: Epoch AI benchmarking hub, read October 22, 2026) - Mystery game puzzles: 8.0% (source: Epoch AI benchmarking hub, read October 22, 2026) - OTIS Mock AIME: 71.1% (source: Epoch AI benchmarking hub, read October 22, 2026) - ProofBench: 18.0% (source: Epoch AI benchmarking hub, read October 22, 2026) - DTBench: 78.9% (source: Epoch AI benchmarking hub, read October 22, 2026) - LMCA: 33.7% (source: Epoch AI benchmarking hub, read October 22, 2026) - Surface Evolver Bench: 55.0% (source: Epoch AI benchmarking hub, read October 22, 2026) - ALE-Bench: 640.0 (source: ALE-Bench (Sakana AI), read October 22, 2026) - Vending-Bench 2: $2,158 (source: Epoch AI benchmarking hub, read October 22, 2026) - Epoch Capabilities Index: 146.9 (source: Epoch AI benchmarking hub, read October 22, 2026) - LMArena Text: 1433 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: coding: 1470 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: hard prompts: 1447 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: math: 1429 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: instruction following: 1433 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: creative writing: 1404 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: 1251 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: English: 1443 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: non-English: 1420 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: Chinese: 1466 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: German: 1428 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: Japanese: 1380 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: Korean: 1373 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: French: 1444 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: Spanish: 1433 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: Russian: 1426 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: Polish: 1423 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: multi-turn: 1443 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: longer queries: 1446 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: expert: 1461 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: hard prompts in English: 1455 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: software and IT: 1466 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: mathematical work: 1448 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: science: 1452 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: medicine and healthcare: 1441 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: legal and government: 1437 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: business and finance: 1434 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: writing and language: 1419 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Text: entertainment and media: 1403 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena WebDev: React: 1478 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena WebDev: HTML: 1455 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena WebDev: from an image: 1470 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: English: 1263 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: Chinese: 1311 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: reading text: 1259 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: diagrams: 1266 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: homework: 1268 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: humor: 1235 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: recognizing things: 1197 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Vision: creative writing: 1234 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026) - LMArena Document: 1435 Elo (source: LMArena (Hugging Face dataset), read October 22, 2026)