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
151Gemini 3.6 Flashmedium· Epoch AI Benchmarking Hub data50.4%
152GLM 5.3 Flashhigh· Epoch AI Benchmarking Hub data50.1%
153Gemini 4 Carbonlow· Epoch AI Benchmarking Hub data49.6%
154GPT-5.6 Lunaxhigh· Epoch AI Benchmarking Hub data47.6%
155Muse Spark Open 816B A42Bhigh· Epoch AI Benchmarking Hub data47.5%
156DeepSeek V4 Flash 0731low· Epoch AI Benchmarking Hub data46%
157Gemini 3 Pro PreviewDeep Thinking· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.45.1%
158GPT 5.2high· Epoch AI Benchmarking Hub data43.3%
159GPT-5.6 Sollow· Epoch AI Benchmarking Hub data42.5%
160Qwen 3.8 27Bxhigh· Epoch AI Benchmarking Hub data42.4%
161Gemini 4 Argonlow· Epoch AI Benchmarking Hub data42.1%
162GPT-6 Lunaxhigh· Epoch AI Benchmarking Hub data41.9%
163Inkling Smallxhigh· Epoch AI Benchmarking Hub data40.1%
164GPT-5.2 Promedium· Epoch AI Benchmarking Hub data38.5%
165Claude Opus 4.564k· Epoch AI Benchmarking Hub data37.6%
166Claude Opus 4.5Opus 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.37.6%
167Claude Opus 4.5private validation · 64K thinking tokens· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.37.6%
168GPT-5.6 Terramedium· Epoch AI Benchmarking Hub data37.5%
169Inkling FP4· Epoch AI Benchmarking Hub data36.5%
170Gemini 3 Flash· Epoch AI Benchmarking Hub data33.6%
171Gemini 3 FlashARC Prize Verified · semi-private set· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.33.6%
172Gemini 3 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.33.6%
173GPT 5.5low· Epoch AI Benchmarking Hub data33.3%
174Grok 4.5low· Epoch AI Benchmarking Hub data33.1%
175Inkling Smallhigh· Epoch AI Benchmarking Hub data33.1%
176Claude Sonnet 5.5medium· Epoch AI Benchmarking Hub data32.8%
177GPT-6 Sollow· Epoch AI Benchmarking Hub data31.5%
178GPT-6 Lunahigh· Epoch AI Benchmarking Hub data31.4%
179Gemini 3 Pro Preview· Epoch AI Benchmarking Hub data31.1%
180Gemini 3 Pro PreviewGemini 3 Pro Model Card · results as of November 2025 · ARC Prize Verified· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.31.1%
181Claude Haiku 5.5high· Epoch AI Benchmarking Hub data30.7%
182Gemini 3.6 Flashlow· Epoch AI Benchmarking Hub data30.4%
183GPT-5.6 Lunahigh· Epoch AI Benchmarking Hub data29.3%
184GPT 5.4low· Epoch AI Benchmarking Hub data29.2%
185GLM 5.3 Flashlow· Epoch AI Benchmarking Hub data27.9%
186Grok 4.6low· Epoch AI Benchmarking Hub data27.6%
187Muse Spark Open 816B A42Bmedium· Epoch AI Benchmarking Hub data27.5%
188GPT 5.2medium· Epoch AI Benchmarking Hub data26.7%
189Claude Sonnet 5.5low· Epoch AI Benchmarking Hub data24%
190Claude Opus 4.516k· Epoch AI Benchmarking Hub data22.8%
191Qwen 3.8 27Blow· Epoch AI Benchmarking Hub data22.8%
192GLM 5.2· Epoch AI Benchmarking Hub data22.8%
193GPT-5.4 minixhigh· Epoch AI Benchmarking Hub data18.9%
194GPT-5.6 Terralow· Epoch AI Benchmarking Hub data18.8%
195GPT-5 Pro· Epoch AI Benchmarking Hub data18.3%
196GPT-6 Lunamedium· Epoch AI Benchmarking Hub data18.1%
197GPT 5.1high· Epoch AI Benchmarking Hub data17.6%
198Grok 4.7low· Epoch AI Benchmarking Hub data16.7%
199Grok 4· Epoch AI Benchmarking Hub data16%
200Claude Opus 4.58k· Epoch AI Benchmarking Hub data13.9%

Showing 151–200 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