Skip to content

Benchmark

GDP.pdf

Scores are percentages. Higher is better.

Models measured
37
Scores recorded
55
Sources
3
IQ weight
0.00

Score distribution

20%
Mean
20.8%
Median
22.1%
Std dev
0.436%
10th percentile
35%
90th percentile
0.328%
Min
0.328%81.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.00 for chat models, measured over 37 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 Sonnet 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim81.6%
2Claude Sonnet 5mean criteria pass rate · Python + image crop · 5 runs· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.81.6%
3Claude Sonnet 4.6Python + image crop · Sonnet 5 card rerun · 128K max output· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.78.6%
4Claude Sonnet 5mean criteria pass rate · no tools · base64 PDF · 5 runs· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.67.5%
5Claude Sonnet 4.6no tools · Sonnet 5 card rerun · 128K max output· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.66.9%
6Gemini 4 Carbonhigh· Epoch AI Benchmarking Hub data56.4%
7Muse Wonder 1.0max· Epoch AI Benchmarking Hub data54.8%
8Gemini 4 Hydrogenxhigh· Epoch AI Benchmarking Hub data54.2%
9Gemini 4 Argonhigh· Epoch AI Benchmarking Hub data53.6%
10Gemini 4 Neonhigh· Epoch AI Benchmarking Hub data53.4%
11Muse Spark 1.3max· Epoch AI Benchmarking Hub data53%
12Claude Fable 5.2max· Epoch AI Benchmarking Hub data51.8%
13GPT-6.1 Solmax· Epoch AI Benchmarking Hub data51.6%
14Claude Sonnet 5.5max· Epoch AI Benchmarking Hub data51.2%
15GPT-6 Astramax· Epoch AI Benchmarking Hub data50.4%
16Claude Fable 5.1max· Epoch AI Benchmarking Hub data49.4%
17GPT-6.2 Astramax· Epoch AI Benchmarking Hub data49.2%
18Muse Ember 1.0max· Epoch AI Benchmarking Hub data49.2%
19Claude Opus 5.5max· Epoch AI Benchmarking Hub data48.8%
20GPT-6.2 Solmax· Epoch AI Benchmarking Hub data48.8%
21Grok 4.7xhigh· Epoch AI Benchmarking Hub data48.8%
22Kimi K3.1 Promax· Epoch AI Benchmarking Hub data46.6%
23Gemini 3.8 Flashhigh· Epoch AI Benchmarking Hub data44.2%
24Inkling 2 Smallxhigh· Epoch AI Benchmarking Hub data43.6%
25Qwen3.8 Maxxhigh· Epoch AI Benchmarking Hub data42.8%
26Kimi K3max· Epoch AI Benchmarking Hub data39.8%
27GLM 5.3max· Epoch AI Benchmarking Hub data36.6%
28Hy4 preview· Epoch AI Benchmarking Hub data35%
29Gemini 3.8 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim35%
30Gemini 3.8 Flashall-pass rate · self-computed· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.35%
31Kimi K3.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim34.6%
32Gemini 3.7 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim34%
33Gemini 3.7 Flashself-computed· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.34%
34Muse Wonder Full· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim33.2%
35Mistral Large 4high· Epoch AI Benchmarking Hub data32.8%
36GPT-6.1 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim32%
37Muse Wonder· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim30.8%
38Muse Wonder 1.0· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim30.8%
39GPT-5.6 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim30.7%
40GPT-5.6 SolOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.30.7%
41Claude Fable 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim29.8%
42Claude Fable 5Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.29.8%
43Claude Fable 5OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.29.8%
44GPT 5.5OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.26%
45GPT 5.5Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.24.9%
46GPT-5.6 Terra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim24.7%
47GPT-5.6 TerraOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.24.7%
48GPT-5.6 Luna· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim22.7%
49GPT-5.6 LunaOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.22.7%
50Claude Opus 4.8Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.22.5%

Showing 1–50 of 55 results.

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats · Epoch AI Benchmarking Hub, CC BY 4.0 (licence)