Skip to content

Benchmark

ARC-AGI-3

Scores are percentages. Higher is better.

Models measured
9
Scores recorded
15
Sources
2
IQ weight
2.66

Score distribution

24.5%
Mean
7.8%
Median
35.8%
Std dev
0.18%
10th percentile
99.9%
90th percentile
0.18%
Min
0.18%99.9%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.66 for chat models, measured over 9 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT-6 Astra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim99.9%
2GPT-6 AstraOpenAI launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.99.9%
3Claude Fable 5.2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim79.3%
4Claude Opus 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim30.2%
5Claude Opus 5OpenAI launch table· Reported by OpenAI, Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.30.2%
6GPT-5.6 SolOpenAI launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.7.8%
7GPT-5.6 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim7.8%
8GPT-5.6 SolOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.7.8%
9Claude Opus 4.8OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1.5%
10GPT-5.6 Terra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim0.8%
11GPT-5.6 TerraOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.0.8%
12GPT 5.5OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.0.43%
13Gemini 3.1 Pro PreviewOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.0.42%
14GPT-5.6 Luna· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim0.18%
15GPT-5.6 LunaOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.0.18%

Data credits: Data by LLM Stats · 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.