Skip to content

Benchmark

ARC-AGI 2

Scores are percentages. Higher is better.

Models measured
98
Scores recorded
318
Sources
3
IQ weight
1.29

Score distribution

42.4%
Mean
42.5%
Median
35%
Std dev
0.9%
10th percentile
90%
90th percentile
0%
Min
0%95.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.29 for chat models, measured over 98 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51GPT-6.2 Solhigh· Epoch AI Benchmarking Hub data85.8%
52GPT-6.2 Solmax· Epoch AI Benchmarking Hub data85.8%
53GPT-6.2 Solxhigh· Epoch AI Benchmarking Hub data85.8%
54GPT-5.6 Solhigh· Epoch AI Benchmarking Hub data85.4%
55GPT-6 Astralow· Epoch AI Benchmarking Hub data85.4%
56GPT 5.5Verified · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.85%
57GPT 5.5xhigh· Epoch AI Benchmarking Hub data85%
58GPT 5.5Gemini 3.5 Flash Model Card · May 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.84.6%
59Gemini 3.7 Flashhigh· Epoch AI Benchmarking Hub data84.6%
60GPT-5.5 Prohigh· Epoch AI Benchmarking Hub data84.6%
61GPT-5.5 Proxhigh· Epoch AI Benchmarking Hub data84.2%
62GPT-5.6 Terramax· Epoch AI Benchmarking Hub data83.9%
63Claude Sonnet 5.5xhigh· Epoch AI Benchmarking Hub data83.9%
64GPT 5.5high· Epoch AI Benchmarking Hub data83.3%
65GPT-5.4 Pro· Epoch AI Benchmarking Hub data83.3%
66GPT-5.4 ProVerified· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.83.3%
67Gemini 3.8 Flashmedium· Epoch AI Benchmarking Hub data82.9%
68GPT-6.2 Solmedium· Epoch AI Benchmarking Hub data82.9%
69Claude Fable 5medium· Epoch AI Benchmarking Hub data82.5%
70Inkling 2 Smallxhigh· Epoch AI Benchmarking Hub data81.7%
71Claude Fable 5.1low· Epoch AI Benchmarking Hub data78.3%
72DeepSeek V4.1 Flash 1017max· Epoch AI Benchmarking Hub data78.1%
73GPT-6 Solxhigh· Epoch AI Benchmarking Hub data78.1%
74Gemini 3.8 Flashlow· Epoch AI Benchmarking Hub data77.5%
75Gemini 3.1 Pro Preview· Epoch AI Benchmarking Hub data77.1%
76Gemini 3.1 Pro PreviewARC Prize Verified· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.77.1%
77Gemini 3.1 Pro PreviewGemini 3.5 Flash Model Card · May 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.77.1%
78Claude Fable 5low· Epoch AI Benchmarking Hub data76.8%
79GPT-6.1 Sollow· Epoch AI Benchmarking Hub data76.7%
80Claude Opus 4.7max· Epoch AI Benchmarking Hub data75.8%
81Claude Opus 4.7private validation · adaptive thinking · max effort · capability summary· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.75.8%
82Claude Opus 4.7Gemini 3.5 Flash Model Card · May 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.75.8%
83GPT-5.6 Terraxhigh· Epoch AI Benchmarking Hub data74.2%
84GPT 5.4xhigh· Epoch AI Benchmarking Hub data74%
85DeepSeek V4.1 Flash 1017low· Epoch AI Benchmarking Hub data73.6%
86GPT 5.4Verified· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.73.3%
87DeepSeek V4.1 Flashmax· Epoch AI Benchmarking Hub data72.9%
88Claude Opus 4.8Opus 5 System Card capability summary · adaptive thinking · max effort unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.72.1%
89Gemini 3.5 FlashGemini 3.5 Flash Model Card · May 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.72.1%
90Claude Opus 4.8high· Epoch AI Benchmarking Hub data72.1%
91Gemini 3.5 Flashhigh· Epoch AI Benchmarking Hub data72.1%
92Claude Opus 4.8medium· Epoch AI Benchmarking Hub data71.7%
93GPT 5.5medium· Epoch AI Benchmarking Hub data70.4%
94Claude Opus 5.5low· Epoch AI Benchmarking Hub data70.1%
95Claude Opus 4.6120k· Epoch AI Benchmarking Hub data69.2%
96Claude Opus 4.6120K thinking tokens · high effort· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.69.2%
97GPT-6 Solhigh· Epoch AI Benchmarking Hub data68.9%
98Claude Opus 4.6Opus 4.6 System Card capability summary · adaptive thinking · max effort · default sampling · 5 trials unless noted · Verified· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.68.8%
99DeepSeek V4.1 Flash 1017high· Epoch AI Benchmarking Hub data68.6%
100Claude Opus 4.7high· Epoch AI Benchmarking Hub data68.3%

Showing 51–100 of 318 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. · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · Data by LLM Stats