Skip to main content

Vals MMMU (MMMU)

Multimodal Multi-task Benchmark

Data verified

How BenchLM shows MMMU

BenchLM mirrors the public Vals AI MMMU leaderboard captured from https://www.vals.ai/benchmarks/mmmu and updated by Vals on July 9, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.

MMMU is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.

81 Vals rows1 task viewspublic datasetTasks: OverallDisplay only

MMMU score on MMMU — July 9, 2026

BenchLM mirrors the published mmmu score view for MMMU. Claude Fable 5 leads the public snapshot at 89.31% , followed by GPT-5.6 Sol (88.84%) and Gemini 3.5 Flash (88.27%). BenchLM does not use these results to rank models overall.

81 modelsExternal benchmark mirrorsCurrentDisplay onlyUpdated July 9, 2026

The published MMMU snapshot is tightly clustered at the top: Claude Fable 5 sits at 89.31%, while the third row is only 1.04 points behind. The broader top-10 spread is 2.64 points, so many of the published scores sit in a relatively narrow band.

81 models have been evaluated on MMMU. The benchmark falls in the External benchmark mirrors category. BenchLM tracks this category separately from its weighted global scoring system, so these results are best compared on the dedicated Korean benchmark views. MMMU is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About MMMU

Year

2026

Tasks

Multimodal academic task suite

Format

Accuracy score

Difficulty

Multimodal college-level reasoning

BenchLM mirrors the public Vals AI MMMU leaderboard as display-only external evidence. The captured snapshot preserves overall scores, task-level scores where Vals publishes them, uncertainty, latency, and cost-per-test metadata. It is excluded from BenchLM weighted rankings.

BenchLM freshness & provenance

Version

MMMU 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

MMMU score table (81 models)

1
Claude Fable 5anthropic/claude-fable-5
89.31%
2
GPT-5.6 Solopenai/gpt-5.6-sol
88.84%
3
Gemini 3.5 Flashgoogle/gemini-3.5-flash
88.27%
4
GPT-5.5openai/gpt-5.5
88.27%
5
Gemini 3.1 Pro Previewgoogle/gemini-3.1-pro-preview
88.21%
6
Gemini 3 Flash Previewgoogle/gemini-3-flash-preview
87.63%
7
Gemini 3 Pro Previewgoogle/gemini-3-pro-preview
87.51%
8
GPT-5.4openai/gpt-5.4-2026-03-05
87.51%
9
Muse Sparkmeta/muse_spark
87.40%
10
GPT-5.2openai/gpt-5.2-2025-12-11
86.67%
11
Muse Spark 1 1meta/muse_spark_1_1
86.59%
12
Claude Opus 4.8anthropic/claude-opus-4-8
86.59%
13
GPT-5.6 Terraopenai/gpt-5.6-terra
86.47%
14
Kimi K2.6kimi/kimi-k2.6
86.30%
15
Claude Opus 4.7anthropic/claude-opus-4-7
85.55%
16
GPT-5.6 Lunaopenai/gpt-5.6-luna
85.03%
17
Kimi K2.5 Thinkingkimi/kimi-k2.5-thinking
84.33%
18
Qwen3.6 Plusalibaba/qwen3.6-plus
84.16%
19
Claude Opus 4.6 Thinkinganthropic/claude-opus-4-6-thinking
83.87%
20
Claude Sonnet 4.6anthropic/claude-sonnet-4-6
83.58%
21
Grok 4.20 0309 Reasoninggrok/grok-4.20-0309-reasoning
83.47%
22
GPT-5.1openai/gpt-5.1-2025-11-13
83.18%
23
Grok 4.3grok/grok-4.3
83.06%
24
Claude Sonnet 5anthropic/claude-sonnet-5
83.01%
25
Claude Opus 4.5 20251101 Thinkinganthropic/claude-opus-4-5-20251101-thinking
82.95%
26
Gemini 3.1 Flash Lite Previewgoogle/gemini-3.1-flash-lite-preview
82.49%
27
Qwen3.5 Flashalibaba/qwen3.5-flash
81.91%
28
GPT-5openai/gpt-5-2025-08-07
81.50%
29
Gemini 2.5 Pro Exp 03 25google/gemini-2.5-pro-exp-03-25
81.34%
30
MiniMax M3minimax/MiniMax-M3
81.16%
31
Claude Opus 4.5anthropic/claude-opus-4-5-20251101
81.10%
32
Gemini 2.5 Flash Preview 09 2025 Thinkinggoogle/gemini-2.5-flash-preview-09-2025-thinking
80.75%
33
O3openai/o3-2025-04-16
80.42%
34
Mimo V2.5xiaomi/mimo-v2.5
80.00%
35
O4 Miniopenai/o4-mini-2025-04-16
79.67%
36
Gemini 2.5 Flash Preview 09 2025google/gemini-2.5-flash-preview-09-2025
79.48%
37
Claude Sonnet 4.5 20250929 Thinkinganthropic/claude-sonnet-4-5-20250929-thinking
79.31%
38
GPT-5.4 Miniopenai/gpt-5.4-mini-2026-03-17
79.25%
39
GPT-5 Miniopenai/gpt-5-mini-2025-08-07
78.91%
40
Claude Opus 4.1 20250805 Thinkinganthropic/claude-opus-4-1-20250805-thinking
77.51%
41
O1openai/o1-2024-12-17
77.41%
42
Grok 4 0709grok/grok-4-0709
76.27%
43
Gemini 2.5 Flash Lite Preview 09 2025 Thinkinggoogle/gemini-2.5-flash-lite-preview-09-2025-thinking
75.43%
44
Claude 3.7 Sonnet 20250219 Thinkinganthropic/claude-3-7-sonnet-20250219-thinking
75.10%
45
Claude Sonnet 4 20250514 Thinkinganthropic/claude-sonnet-4-20250514-thinking
74.93%
46
Claude Opus 4.1anthropic/claude-opus-4-1-20250805
73.72%
47
GPT-5.4 Nanoopenai/gpt-5.4-nano-2026-03-17
73.58%
48
Claude Opus 4anthropic/claude-opus-4-20250514
73.31%
49
Grok 4 Fast Reasoninggrok/grok-4-fast-reasoning
72.78%
50
Grok 4.1 Fast Reasoninggrok/grok-4-1-fast-reasoning
72.66%
51
Gemini 2.5 Flash Lite Preview 09 2025google/gemini-2.5-flash-lite-preview-09-2025
72.54%
52
Claude Sonnet 4anthropic/claude-sonnet-4-20250514
72.39%
53
GPT-4.1openai/gpt-4.1-2025-04-14
72.39%
54
Gemini 2.5 Flash Preview 04 17 Thinkinggoogle/gemini-2.5-flash-preview-04-17-thinking
71.92%
55
Llama4 Maverick Instruct Basicfireworks/llama4-maverick-instruct-basic
71.69%
56
Claude 3.7 Sonnetanthropic/claude-3-7-sonnet-20250219
71.52%
57
GPT-5 Nanoopenai/gpt-5-nano-2025-08-07
70.94%
58
Command A Plus 05 2026cohere/command-a-plus-05-2026
70.64%
59
GPT-4.1 Miniopenai/gpt-4.1-mini-2025-04-14
70.54%
60
Gemini 2.0 Flash 001google/gemini-2.0-flash-001
69.79%
61
Claude 3.5 Sonnetanthropic/claude-3-5-sonnet-20241022
68.80%
62
Gemini 2.5 Flash Preview 04 17google/gemini-2.5-flash-preview-04-17
68.00%
63
Mistral Large 2512mistralai/mistral-large-2512
66.19%
64
Gemini 1.5 Pro 002google/gemini-1.5-pro-002
65.51%
65
Magistral Small 2509mistralai/magistral-small-2509
65.20%
66
Magistral Medium 2509mistralai/magistral-medium-2509
64.57%
67
GPT-4oopenai/gpt-4o-2024-08-06
64.01%
68
Grok 4.1 Fast Non Reasoninggrok/grok-4-1-fast-non-reasoning
63.70%
69
Grok 4 Fast Non Reasoninggrok/grok-4-fast-non-reasoning
63.41%
70
Mistral Medium 2505mistralai/mistral-medium-2505
62.97%
71
GPT-4oopenai/gpt-4o-2024-11-20
62.16%
72
Mistral Small 2503mistralai/mistral-small-2503
60.08%
73
Meta Llama Llama 4 Scout 17B 16E Instructtogether/meta-llama/Llama-4-Scout-17B-16E-Instruct
58.75%
74
Grok 2 Vision 1212grok/grok-2-vision-1212
57.25%
75
Gemini 1.5 Flash 002google/gemini-1.5-flash-002
57.19%
76
GPT-4o Miniopenai/gpt-4o-mini-2024-07-18
56.56%
77
GPT-4.1 Nanoopenai/gpt-4.1-nano-2025-04-14
55.05%
78
Meta Llama Llama 3.2 90B Vision Instruct Turbotogether/meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo
48.06%
79
Claude Haiku 4.5 20251001 Thinkinganthropic/claude-haiku-4-5-20251001-thinking
46.07%
80
Meta Llama Llama 3.2 11B Vision Instruct Turbotogether/meta-llama/Llama-3.2-11B-Vision-Instruct-Turbo
38.82%
81
Qwen3.5 Plus Thinkingalibaba/qwen3.5-plus-thinking
22.77%

FAQ

What does MMMU measure?

Multimodal Multi-task Benchmark

Which model leads the published MMMU snapshot?

Claude Fable 5 currently leads the published MMMU snapshot with 89.31% mmmu score. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on MMMU?

81 AI models are included in BenchLM's mirrored MMMU snapshot, based on the public leaderboard captured on July 9, 2026.

Last updated: July 9, 2026 · mirrored from the public benchmark leaderboard

The AI models change fast. We track them for you.

For engineers, researchers, and the plain curious — a weekly brief on new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.