Skip to content

Benchmark

Mystery Game Puzzles

Scores are percentages.

Models measured
85
Scores recorded
146
Sources
1
IQ weight
2.37

Score distribution

25.3%
Mean
19%
Median
19.7%
Std dev
7%
10th percentile
56%
90th percentile
0%
Min
0%95%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.37 for chat models, measured over 85 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT-6.2 Astramax· Epoch AI Benchmarking Hub data95%
2Claude Fable 5.2max· Epoch AI Benchmarking Hub data90%
3Gemini 4 Neonxhigh· Epoch AI Benchmarking Hub data85%
4GPT-6 Astramax· Epoch AI Benchmarking Hub data84%
5GPT-6.1 Solmax· Epoch AI Benchmarking Hub data80%
6GPT-6.1 Terramax· Epoch AI Benchmarking Hub data74%
7Gemini 4 Carbonhigh· Epoch AI Benchmarking Hub data73%
8Claude Opus 5.5max· Epoch AI Benchmarking Hub data71%
9Claude Sonnet 5.5max· Epoch AI Benchmarking Hub data65%
10Gemini 4 Hydrogenxhigh· Epoch AI Benchmarking Hub data65%
11Gemini 4 Argonhigh· Epoch AI Benchmarking Hub data64%
12Claude Opus 5max· Epoch AI Benchmarking Hub data59%
13Claude Fable 5.1max· Epoch AI Benchmarking Hub data58%
14GPT-5.6 Solmax· Epoch AI Benchmarking Hub data58%
15GPT 5.5xhigh· Epoch AI Benchmarking Hub data56%
16GPT-6 Solmax· Epoch AI Benchmarking Hub data56%
17Claude Fable 5max· Epoch AI Benchmarking Hub data52%
18GPT 5.5high· Epoch AI Benchmarking Hub data52%
19Muse Wonder 1.0max· Epoch AI Benchmarking Hub data51%
20DeepSeek V4.1 Flash 1017max· Epoch AI Benchmarking Hub data49%
21Gemini 3.8 Flashhigh· Epoch AI Benchmarking Hub data47%
22DeepSeek V4 Pro 0813max· Epoch AI Benchmarking Hub data43%
23DeepSeek V4.1 Flashmax· Epoch AI Benchmarking Hub data43%
24Qwen3.8 Maxxhigh· Epoch AI Benchmarking Hub data38%
25Claude Opus 5· Epoch AI Benchmarking Hub data37%
26Gemini 3.7 Flashhigh· Epoch AI Benchmarking Hub data37%
27GPT 5.4xhigh· Epoch AI Benchmarking Hub data37%
28Claude Opus 4.8max· Epoch AI Benchmarking Hub data36%
29Claude Sonnet 5max· Epoch AI Benchmarking Hub data35%
30GPT-5.6 Terramax· Epoch AI Benchmarking Hub data35%
31DeepSeek V4 Flash 0731max· Epoch AI Benchmarking Hub data34%
32Gemini 3.1 Pro Previewhigh· Epoch AI Benchmarking Hub data34%
33Grok 4.6xhigh· Epoch AI Benchmarking Hub data34%
34GLM 5.3max· Epoch AI Benchmarking Hub data33%
35GPT-5.6 Solnone· Epoch AI Benchmarking Hub data33%
36Gemini 3.1 Pro Previewmedium· Epoch AI Benchmarking Hub data32%
37Gemini 3.5 Flashhigh· Epoch AI Benchmarking Hub data32%
38Qwen3.7 Max· Epoch AI Benchmarking Hub data32%
39Claude Opus 4.8xhigh· Epoch AI Benchmarking Hub data31%
40Claude Haiku 5.5max· Epoch AI Benchmarking Hub data30%
41Gemini 3.6 Flashhigh· Epoch AI Benchmarking Hub data30%
42Gemini 3.1 Pro Previewlow· Epoch AI Benchmarking Hub data29%
43Grok 4.7xhigh· Epoch AI Benchmarking Hub data29%
44OpenAI o3high· Epoch AI Benchmarking Hub data29%
45Claude Opus 4.7max· Epoch AI Benchmarking Hub data28%
46Gemini 3.5 Flashlow· Epoch AI Benchmarking Hub data28%
47GPT 5.4medium· Epoch AI Benchmarking Hub data28%
48GPT 5.5low· Epoch AI Benchmarking Hub data28%
49Gemini 3 Flashlow· Epoch AI Benchmarking Hub data26%
50GPT-5.6 Sollow· Epoch AI Benchmarking Hub data26%

Showing 1–50 of 146 results.

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)