# Massive Multi-discipline Multimodal Understanding Pro (MMMU-Pro) > A harder multimodal benchmark for frontier models that combines text with images, diagrams, charts, and academic visual reasoning tasks. Canonical page: https://benchlm.ai/benchmarks/mmmu-pro - Category: [Multimodal & Grounded](/multimodal-grounded) - Last updated: September 15, 2026 ## About MMMU-Pro - Year: 2024 - Tasks: Multimodal academic reasoning - Format: Image + text question answering - Difficulty: Frontier multimodal - Paper: [MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark](https://arxiv.org/abs/2409.02813) MMMU-Pro extends the original MMMU setup with more difficult multimodal questions and stronger separation at the top end of the model market. MMMU-Pro is currently weighted in BenchLM's scoring formula. The Multimodal & Grounded category carries 12% of the overall score, and MMMU-Pro contributes 40% of that category score. ## Leaderboard (40 models) | Rank | Model | Creator | Score | |------|-------|---------|-------| | 1 | [Gemini 3.1 Pro](/models/gemini-3-1-pro) | Google | 83.9% | | 2 | [Gemini 3.5 Flash](/models/gemini-3-5-flash) | Google | 83.6% | | 3 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | OpenAI | 83% | | 4 | [Qwen3.8 Max](/models/qwen3-8-max) | Alibaba | 82.3% | | 5 | [Kimi K3](/models/kimi-k3) | Moonshot AI | 81.6% | | 6 | [Seed 2.1 Pro](/models/seed-2-1-pro) | ByteDance | 81.6% | | 7 | [GPT-5.5](/models/gpt-5-5) | OpenAI | 81.2% | | 8 | [GPT-5.4](/models/gpt-5-4) | OpenAI | 81.2% | | 9 | [Gemini 3 Pro](/models/gemini-3-pro) | Google | 81% | | 10 | [GPT-5.6 Terra](/models/gpt-5-6-terra) | OpenAI | 80.7% | | 11 | [Muse Spark](/models/muse-spark) | Meta | 80.4% | | 12 | [Seed 2.1 Turbo](/models/seed-2-1-turbo) | ByteDance | 80.1% | | 13 | [GPT-5.2](/models/gpt-5-2) | OpenAI | 79.5% | | 14 | [Kimi K2.6](/models/kimi-2-6) | Moonshot AI | 79.4% | | 15 | [dots3-note Preview](/models/dots3-note-preview) | Dots Studio | 79.1% | | 16 | [Qwen3.7 Plus](/models/qwen3-7-plus) | Alibaba | 79% | | 17 | [Qwen3.5 397B](/models/qwen3-5-397b) | Alibaba | 79% | | 18 | [Qwen3.6 Plus](/models/qwen3-6-plus) | Alibaba | 78.8% | | 19 | [Kimi K2.5 (Reasoning)](/models/kimi-k2-5-reasoning) | Moonshot AI | 78.5% | | 20 | [Kimi K2.5](/models/kimi-k2-5) | Moonshot AI | 78.5% | | 21 | [GPT-5.6 Luna](/models/gpt-5-6-luna) | OpenAI | 78.4% | | 22 | [MiniMax M3](/models/minimax-m3) | MiniMax | 78.1% | | 23 | [Grok 4.3](/models/grok-4-3) | xAI | 78.1% | | 24 | [MiMo-V2.5](/models/mimo-v2-5) | Xiaomi | 77.9% | | 25 | [Claude Opus 4.6](/models/claude-opus-4-6) | Anthropic | 77.3% | | 26 | [Gemma 4 31B](/models/gemma-4-31b) | Google | 76.9% | | 27 | [GPT-5.4 mini](/models/gpt-5-4-mini) | OpenAI | 76.6% | | 28 | [Qwen3.6-27B](/models/qwen3-6-27b) | Alibaba | 75.8% | | 29 | [Qwen3.6-35B-A3B](/models/qwen3-6-35b-a3b) | Alibaba | 75.3% | | 30 | [Grok 4.20](/models/grok-4-20-beta) | xAI | 75.2% | | 31 | [Inkling-Small](/models/inkling-small) | Thinking Machines Lab | 74% | | 32 | [Muse Glimmer 30B](/models/muse-glimmer-30b) | Meta | 74% | | 33 | [Gemma 4 26B A4B](/models/gemma-4-26b-a4b) | Google | 73.8% | | 34 | [Inkling](/models/inkling) | Thinking Machines Lab | 73.5% | | 35 | [Interfaze Beta](/models/interfaze-beta) | Interfaze | 71.1% | | 36 | [Claude Opus 4.5](/models/claude-opus-4-5) | Anthropic | 70.6% | | 37 | [Gemma 4 12B](/models/gemma-4-12b) | Google | 69.1% | | 38 | [GPT-5.4 nano](/models/gpt-5-4-nano) | OpenAI | 66.1% | | 39 | [Command A+](/models/command-a-plus) | Cohere | 63% | | 40 | [LFM2.5-VL-3B](/models/lfm2-5-vl-3b) | LiquidAI | 30.5% | ## FAQ ### What does MMMU-Pro measure? A harder multimodal benchmark for frontier models that combines text with images, diagrams, charts, and academic visual reasoning tasks. ### Which model scores highest on MMMU-Pro? Gemini 3.1 Pro by Google currently leads with a score of 83.9% on MMMU-Pro. ### How many models are evaluated on MMMU-Pro? 40 AI models have been evaluated on MMMU-Pro on BenchLM. ### Does MMMU-Pro affect BenchLM's overall score? Yes. MMMU-Pro is a weighted benchmark inside the Multimodal & Grounded category, which carries 12% of BenchLM's overall score. MMMU-Pro itself contributes 40% of that category score. ## Compare Top Models on MMMU-Pro - [Gemini 3.1 Pro vs Gemini 3.5 Flash](/compare/gemini-3-1-pro-vs-gemini-3-5-flash) - [Gemini 3.5 Flash vs GPT-5.6 Sol](/compare/gemini-3-5-flash-vs-gpt-5-6-sol) - [GPT-5.6 Sol vs Qwen3.8 Max](/compare/gpt-5-6-sol-vs-qwen3-8-max) - [Qwen3.8 Max vs Kimi K3](/compare/kimi-k3-vs-qwen3-8-max)