Skip to content

Benchmark

ARC-AGI

Scores are percentages.

Models measured
102
Scores recorded
295
Sources
2
IQ weight
2.26

Score distribution

64.1%
Mean
75%
Median
31.2%
Std dev
14.5%
10th percentile
97%
90th percentile
0%
Min
0%98.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.26 for chat models, measured over 102 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
251Claude Sonnet 4· Epoch AI Benchmarking Hub data23.8%
252O1 Prolow· Epoch AI Benchmarking Hub data23.3%
253Claude Opus 4· Epoch AI Benchmarking Hub data22.5%
254o3 Minimedium· Epoch AI Benchmarking Hub data22.3%
255o4 Minilow· Epoch AI Benchmarking Hub data21.3%
256DeepSeek R1· Epoch AI Benchmarking Hub data21.2%
257Claude Sonnet 3.78k· Epoch AI Benchmarking Hub data21.2%
258GPT-5 nanomedium· Epoch AI Benchmarking Hub data20.7%
259GPT 5.4 Nanolow· Epoch AI Benchmarking Hub data18.3%
260O1 Preview· Epoch AI Benchmarking Hub data18%
261Gemini 3.5 Flash Litelow· Epoch AI Benchmarking Hub data17%
262Claude Haiku 4.51k· Epoch AI Benchmarking Hub data16.8%
263GPT-5 nanohigh· Epoch AI Benchmarking Hub data16.7%
264Grok 3 Minilow· Epoch AI Benchmarking Hub data16.5%
265Grok 3 Minilow, beta· Epoch AI Benchmarking Hub data16.5%
266Gemini 2.5 Flash Preview 05-20 (no think)1k· Epoch AI Benchmarking Hub data16%
267o3 Minilow· Epoch AI Benchmarking Hub data14.5%
268Claude Haiku 4.5· Epoch AI Benchmarking Hub data14.3%
269O1 Mini· Epoch AI Benchmarking Hub data14%
270Claude Sonnet 3.7· Epoch AI Benchmarking Hub data13.6%
271DeepSeek V4 Pro 0813none· Epoch AI Benchmarking Hub data13%
272GPT-5.4 minilow· Epoch AI Benchmarking Hub data13%
273GPT 5.2none· Epoch AI Benchmarking Hub data12.3%
274DeepSeek V4 Flash 0731none· Epoch AI Benchmarking Hub data11.8%
275Claude Sonnet 3.71k· Epoch AI Benchmarking Hub data11.6%
276GPT-4.5 Preview· Epoch AI Benchmarking Hub data10.3%
277GPT-6 Lunanone· Epoch AI Benchmarking Hub data8.8%
278DeepSeek V4.1 Flashnone· Epoch AI Benchmarking Hub data7.8%
279Gemini 3.5 Flash Liteminimal· Epoch AI Benchmarking Hub data7.5%
280Magistral Medium2506· Epoch AI Benchmarking Hub data6.1%
281GPT 5minimal· Epoch AI Benchmarking Hub data6%
282GPT 5.1none· Epoch AI Benchmarking Hub data5.8%
283DeepSeek V4.1 Flash 1017none· Epoch AI Benchmarking Hub data5.7%
284GPT 4.1· Epoch AI Benchmarking Hub data5.5%
285Grok 3· Epoch AI Benchmarking Hub data5.5%
286Inkling Smallnone· Epoch AI Benchmarking Hub data5.5%
287GPT-5 miniminimal· Epoch AI Benchmarking Hub data5.3%
288Magistral Small 25092506· Epoch AI Benchmarking Hub data5%
289GPT 4o· Epoch AI Benchmarking Hub data4.5%
290Llama 4 Maverick· Epoch AI Benchmarking Hub data4.4%
291GPT-5 nanolow· Epoch AI Benchmarking Hub data4%
292GPT-4.1 mini· Epoch AI Benchmarking Hub data3.5%
293GPT-5 nanominimal· Epoch AI Benchmarking Hub data1.5%
294Llama 4 Scout· Epoch AI Benchmarking Hub data0.5%
295GPT-4.1 nano· Epoch AI Benchmarking Hub data0%

Showing 251–295 of 295 results.

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · Data by LLM Stats