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
1Claude Fable 5max· Epoch AI Benchmarking Hub data98.5%
2Claude Fable 5xhigh· Epoch AI Benchmarking Hub data98.5%
3Claude Fable 5.2xhigh· Epoch AI Benchmarking Hub data98.5%
4Claude Opus 5.5high· Epoch AI Benchmarking Hub data98.5%
5Gemini 3.8 Flashhigh· Epoch AI Benchmarking Hub data98.5%
6Gemini 4 Argonhigh· Epoch AI Benchmarking Hub data98.5%
7Gemini 4 Argonmedium· Epoch AI Benchmarking Hub data98.5%
8Gemini 4 Carbonmedium· Epoch AI Benchmarking Hub data98.5%
9GPT-6 Astra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim98.5%
10GPT-6 Astrahigh· Epoch AI Benchmarking Hub data98.5%
11GPT-6 Astraxhigh· Epoch AI Benchmarking Hub data98.5%
12GPT-6.1 Solhigh· Epoch AI Benchmarking Hub data98.5%
13GPT-6.1 Solxhigh· Epoch AI Benchmarking Hub data98.5%
14GPT-6.2 Astra· Epoch AI Benchmarking Hub data98.5%
15Gemini 3.1 Pro Preview· Epoch AI Benchmarking Hub data98%
16Claude Fable 5.1max· Epoch AI Benchmarking Hub data97.5%
17Claude Fable 5.2high· Epoch AI Benchmarking Hub data97.5%
18Claude Fable 5.2max· Epoch AI Benchmarking Hub data97.5%
19Claude Fable 5.2medium· Epoch AI Benchmarking Hub data97.5%
20Claude Opus 5· Epoch AI Benchmarking Hub data97.5%
21Claude Opus 5.5max· Epoch AI Benchmarking Hub data97.5%
22Claude Opus 5.5medium· Epoch AI Benchmarking Hub data97.5%
23Claude Opus 5.5xhigh· Epoch AI Benchmarking Hub data97.5%
24Gemini 3.8 Flashmedium· Epoch AI Benchmarking Hub data97.5%
25Gemini 4 Carbonhigh· Epoch AI Benchmarking Hub data97.5%
26GPT-5.6 Solxhigh· Epoch AI Benchmarking Hub data97.5%
27GPT-6 Astramax· Epoch AI Benchmarking Hub data97.5%
28GPT-6 Astramedium· Epoch AI Benchmarking Hub data97.5%
29GPT-6.2 Solhigh· Epoch AI Benchmarking Hub data97.5%
30GPT-6.2 Solmax· Epoch AI Benchmarking Hub data97.5%
31GPT-5.6 Solhigh· Epoch AI Benchmarking Hub data97%
32GPT-6.2 Solxhigh· Epoch AI Benchmarking Hub data97%
33Claude Fable 5.1xhigh· Epoch AI Benchmarking Hub data96.5%
34Claude Sonnet 5.5max· Epoch AI Benchmarking Hub data96.5%
35Claude Sonnet 5.5xhigh· Epoch AI Benchmarking Hub data96.5%
36GPT-5.5 Prohigh· Epoch AI Benchmarking Hub data96.5%
37GPT-5.6 Solmax· Epoch AI Benchmarking Hub data96.5%
38GPT-5.6 Terramax· Epoch AI Benchmarking Hub data96.5%
39GPT-6 Astralow· Epoch AI Benchmarking Hub data96.5%
40GPT-6.1 Solmax· Epoch AI Benchmarking Hub data96.5%
41Inkling 2 Smallxhigh· Epoch AI Benchmarking Hub data96.5%
42Claude Fable 5.1high· Epoch AI Benchmarking Hub data96%
43Claude Fable 5high· Epoch AI Benchmarking Hub data95.5%
44Gemini 3.7 Flashhigh· Epoch AI Benchmarking Hub data95.5%
45GPT-6 Solmax· Epoch AI Benchmarking Hub data95.5%
46GPT-6.1 Solmedium· Epoch AI Benchmarking Hub data95.5%
47GPT 5.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim95%
48GPT 5.5xhigh· Epoch AI Benchmarking Hub data95%
49GPT-5.5 Proxhigh· Epoch AI Benchmarking Hub data95%
50Claude Fable 5.1medium· Epoch AI Benchmarking Hub data94.5%

Showing 1–50 of 295 results.

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