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
201Inkling Smallminimal· Epoch AI Benchmarking Hub data48.5%
202Claude Sonnet 4.516k· Epoch AI Benchmarking Hub data48.3%
203Claude Haiku 4.532k· Epoch AI Benchmarking Hub data47.7%
204GLM 5.3 Flashlow· Epoch AI Benchmarking Hub data47%
205Claude Sonnet 4.58k· Epoch AI Benchmarking Hub data46.5%
206GLM 5· Epoch AI Benchmarking Hub data44.7%
207O3 Prolow· Epoch AI Benchmarking Hub data44.3%
208GPT 5low· Epoch AI Benchmarking Hub data44%
209o4 Minimedium· Epoch AI Benchmarking Hub data41.8%
210Qwen3 235B A22B Instruct 2507· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
41.8%
211OpenAI o3low· Epoch AI Benchmarking Hub data41.5%
212Gemini 2.5 Pro16k· Epoch AI Benchmarking Hub data41%
213GPT-5.4 minimedium· Epoch AI Benchmarking Hub data40.8%
214Claude Opus 4.5· Epoch AI Benchmarking Hub data40%
215Claude Sonnet 416k· Epoch AI Benchmarking Hub data40%
216GPT 5.4 Nanohigh· Epoch AI Benchmarking Hub data38.2%
217GPT-6 Lunalow· Epoch AI Benchmarking Hub data37.7%
218Claude Haiku 4.516k· Epoch AI Benchmarking Hub data37.3%
219GPT-5 minimedium· Epoch AI Benchmarking Hub data37.3%
220Gemini 2.5 Pro32k· Epoch AI Benchmarking Hub data37%
221Claude Opus 416k· Epoch AI Benchmarking Hub data35.7%
222Gemini 3.6 Flashminimal· Epoch AI Benchmarking Hub data34.5%
223o3 Minihigh· Epoch AI Benchmarking Hub data34.5%
224GPT-5.6 Lunalow· Epoch AI Benchmarking Hub data34.2%
225Qwen 3.8 27Bnone· Epoch AI Benchmarking Hub data34%
226Gemini 2.5 Flash Preview 05-20 (no think)· Epoch AI Benchmarking Hub data33.3%
227Gemini 2.5 Flash Preview 05-20 (no think)16k· Epoch AI Benchmarking Hub data33.3%
228GPT 5.1low· Epoch AI Benchmarking Hub data33.2%
229Gemini 2.5 Pro Preview 03-25· Epoch AI Benchmarking Hub data33%
230GPT 5.4 Nanomedium· Epoch AI Benchmarking Hub data33%
231Gemini 3.5 Flash Litemedium· Epoch AI Benchmarking Hub data32.3%
232Gemini 2.5 Flash Preview 05-20 (no think)23k· Epoch AI Benchmarking Hub data32.3%
233gemini-2.5-flash-preview-04-1724k· Epoch AI Benchmarking Hub data32.3%
234Gemini 2.5 Pro1k· Epoch AI Benchmarking Hub data31.3%
235Claude Sonnet 4.51k· Epoch AI Benchmarking Hub data31%
236o1medium· Epoch AI Benchmarking Hub data30.7%
237Claude Opus 48k· Epoch AI Benchmarking Hub data30.7%
238Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim29.8%
239Gemini 2.5 Pro8k· Epoch AI Benchmarking Hub data29.5%
240GPT-6 Solnone· Epoch AI Benchmarking Hub data29.3%
241Claude Sonnet 48k· Epoch AI Benchmarking Hub data29%
242Claude Sonnet 3.716k· Epoch AI Benchmarking Hub data28.6%
243Claude Sonnet 41k· Epoch AI Benchmarking Hub data28%
244Codex Mini· Epoch AI Benchmarking Hub data27.3%
245o1low· Epoch AI Benchmarking Hub data27.2%
246Claude Opus 41k· Epoch AI Benchmarking Hub data27%
247GPT-5 minilow· Epoch AI Benchmarking Hub data26.3%
248Gemini 2.5 Flash Preview 05-20 (no think)8k· Epoch AI Benchmarking Hub data25.8%
249Claude Haiku 4.58k· Epoch AI Benchmarking Hub data25.5%
250Claude Sonnet 4.5· Epoch AI Benchmarking Hub data25.5%

Showing 201–250 of 295 results.

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