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
201Inkling Smallmedium· Epoch AI Benchmarking Hub data13.6%
202Claude Sonnet 4.532k· Epoch AI Benchmarking Hub data13.6%
203Claude Sonnet 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.13.6%
204Qwen 3.8 27Bmedium· Epoch AI Benchmarking Hub data13.2%
205GPT-5.4 minihigh· Epoch AI Benchmarking Hub data13.2%
206Kimi K3low· Epoch AI Benchmarking Hub data12.4%
207Muse Spark Open 816B A42Blow· Epoch AI Benchmarking Hub data12.2%
208Kimi K2.5· Epoch AI Benchmarking Hub data11.8%
209Gemini 3.5 Flash Litehigh· Epoch AI Benchmarking Hub data10.3%
210GPT 5high· Epoch AI Benchmarking Hub data9.9%
211GPT 5.2low· Epoch AI Benchmarking Hub data9.7%
212Claude Haiku 5.5medium· Epoch AI Benchmarking Hub data9.4%
213Gemini 3.5 Flashminimal· Epoch AI Benchmarking Hub data8.9%
214Claude Opus 416k· Epoch AI Benchmarking Hub data8.6%
215Claude Opus 4.5· Epoch AI Benchmarking Hub data7.8%
216GPT 5medium· Epoch AI Benchmarking Hub data7.5%
217GPT-5.6 Lunamedium· Epoch AI Benchmarking Hub data7.4%
218Claude Sonnet 4.516k· Epoch AI Benchmarking Hub data6.9%
219Claude Sonnet 4.58k· Epoch AI Benchmarking Hub data6.9%
220GPT 5.1medium· Epoch AI Benchmarking Hub data6.5%
221OpenAI o3high· Epoch AI Benchmarking Hub data6.5%
222o4 Minihigh· Epoch AI Benchmarking Hub data6.1%
223Claude Sonnet 416k· Epoch AI Benchmarking Hub data5.9%
224Claude Sonnet 4.51k· Epoch AI Benchmarking Hub data5.8%
225GPT 5.4 Nanoxhigh· Epoch AI Benchmarking Hub data5.7%
226Grok 4 Fast· Epoch AI Benchmarking Hub data5.3%
227Gemini 3.5 Flash Litemedium· Epoch AI Benchmarking Hub data5.3%
228GPT-5.6 Lunalow· Epoch AI Benchmarking Hub data5.1%
229Inkling Smalllow· Epoch AI Benchmarking Hub data4.9%
230Gemini 2.5 Pro32k· Epoch AI Benchmarking Hub data4.9%
231GLM 5· Epoch AI Benchmarking Hub data4.9%
232MiniMax M2.5· Epoch AI Benchmarking Hub data4.9%
233O3 Prohigh· Epoch AI Benchmarking Hub data4.9%
234GPT-6 Lunalow· Epoch AI Benchmarking Hub data4.6%
235Claude Opus 48k· Epoch AI Benchmarking Hub data4.5%
236GPT-5 minihigh· Epoch AI Benchmarking Hub data4.4%
237GPT-5.4 minimedium· Epoch AI Benchmarking Hub data4.4%
238Claude Haiku 4.532k· Epoch AI Benchmarking Hub data4%
239DeepSeek V3.2· Epoch AI Benchmarking Hub data4%
240Gemini 2.5 Pro16k· Epoch AI Benchmarking Hub data4%
241GPT-5 minimedium· Epoch AI Benchmarking Hub data4%
242Inkling Smallminimal· Epoch AI Benchmarking Hub data4%
243Claude Sonnet 4.5· Epoch AI Benchmarking Hub data3.8%
244GPT 5.4 Nanohigh· Epoch AI Benchmarking Hub data3.6%
245o3 Minihigh· Epoch AI Benchmarking Hub data3%
246OpenAI o3medium· Epoch AI Benchmarking Hub data3%
247Gemini 2.5 Pro8k· Epoch AI Benchmarking Hub data2.9%
248Claude Haiku 4.516k· Epoch AI Benchmarking Hub data2.8%
249Claude Haiku 5.5low· Epoch AI Benchmarking Hub data2.8%
250Gemini 3.6 Flashminimal· Epoch AI Benchmarking Hub data2.6%

Showing 201–250 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