Skip to content

Benchmark

Vals MMMU Pro

Vision benchmarks measure how well models understand images. Scores are percentages.

Models measured
83
Scores recorded
88
Sources
1
IQ weight
1.88

Score distribution

76.3%
Mean
80%
Median
12.4%
Std dev
60.1%
10th percentile
88.2%
90th percentile
22.8%
Min
22.8%90.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.88 for chat models, measured over 83 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude Fable 5.1· Vals AI90.6%
2Claude Opus 5· Vals AI89.9%
3Claude Fable 5· Vals AI89.3%
4Gemini 3.8 Flash· Vals AI89.1%
5Gemini 3.7 Flash· Vals AI89%
6GPT-5.6 Sol· Vals AI88.8%
7Gemini 3.6 Flash· Vals AI88.4%
8Gemini 3.5 Flash· Vals AI88.3%
9GPT 5.5· Vals AI88.3%
10Gemini 3.1 Pro Preview· Vals AI88.2%
11Kimi K3· Vals AI88.2%
12Qwen3.8 Max· Vals AI88%
13Gemini 3 Flash· Vals AI87.6%
14Gemini 3 Pro Preview· Vals AI87.5%
15GPT 5.4· Vals AI87.5%
16Muse Spark· Vals AI87.4%
17GPT 5.2· Vals AI86.7%
18Claude Opus 4.8· Vals AI86.6%
19Muse Spark 1.1· Vals AI86.6%
20GPT-5.6 Terra· Vals AI86.5%
21Kimi K2.6· Vals AI86.3%
22Muse Spark 1.2· Vals AI86.1%
23GLM 5.3 Flash· Vals AI86%
24Claude Opus 4.7· Vals AI85.5%
25GPT-5.6 Luna· Vals AI85%
26Kimi K2.5thinking· Vals AI84.3%
27Qwen3.6 Plus· Vals AI84.2%
28Qwen 3.8 27B· Vals AI83.9%
29Claude Opus 4.6· Vals AI83.9%
30Claude Sonnet 4.6· Vals AI83.6%
31Gemini 3.5 Flash Lite· Vals AI83.6%
32Grok 4.20· Vals AI83.5%
33GPT 5.1· Vals AI83.2%
34Grok 4.3· Vals AI83.1%
35Claude Sonnet 5· Vals AI83%
36Claude Opus 4.5· Vals AI82.9%
37Gemini 3.1 Flash Lite· Vals AI82.5%
38Qwen3.5 Flash· Vals AI81.9%
39GPT 5· Vals AI81.5%
40Gemini 2.5 Pro· Vals AI81.3%
41MiniMax M3· Vals AI81.2%
42gemini-2.5-flash-preview-09-2025thinking· Vals AI80.8%
43OpenAI o3· Vals AI80.4%
44MiMo V2.5· Vals AI80%
45o4 Mini· Vals AI79.7%
46gemini-2.5-flash-preview-09-2025· Vals AI79.5%
47Claude Sonnet 4.5· Vals AI79.3%
48GPT-5.4 mini· Vals AI79.2%
49GPT-5 mini· Vals AI78.9%
50Inkling FP4· Vals AI78.5%

Showing 1–50 of 88 results.