Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Abstract Reasoning on ARC-AGI v2 (test)
Loading...
100
Accuracy
Best Human
-4
23
50
77
May 19, 2026
Accuracy
Updated 14d ago
Evaluation Results
Method
Method
Links
Accuracy
Best Human
2026.05
100
Gemini 3 Pro
#Params=N/A
2026.05
31.1
Grok-4-thinking
#Params=1.7T
2026.05
16
GRAM
#Params=10M, Supervisi...
2026.05
11.1
GPT 5.2 (low)
#Params=N/A
2026.05
9.7
TRM
#Params=7M, Supervisio...
2026.05
7.8
HRM
#Params=27M, Supervisi...
2026.05
5
o3-mini-high
#Params=N/A
2026.05
3
Deepseek-R1
#Params=671B
2026.05
1.3
Claude 3.7 16k
#Params=N/A
2026.05
0.7
Direct Pred
#Params=27M, Supervisi...
2026.05
0
Feedback
Search any
task
Search any
task