Reasoning
MATH
accuracy on the original competition mathematics test set
Rankings
Higher is better#191.8%
Gemini 2.0 Pro
289%
Gemma 3OW
386.5%
Gemini 1.5 Pro-002
477.9%
Gemini 1.5 Flash-002
577%
LLaMA 3.3OW
676.6%
GPT-4o
773.8%
LLaMA 3.1
870.2%
GPT-4o mini
968%
LLaMA 3.2OW
1064.3%
GPT-4 Turbo
1158.5%
Gemini 1.5 Pro
1254.9%
Gemini 1.5 Flash
1353.2%
Gemini 1.0 Ultra
1450.4%
LLaMA 3 (8B/70B)
1542.3%
Gemma 2OW
1635.9%
Gemini 1.5 Flash-8B
1732.6%
Gemini 1.0 Pro
1824.3%
Gemma 1OW
1922.8%
Gemini 1.0 Nano
2013.5%
LLaMA 2
2110.6%
LLaMA 1
MATH — frequently asked questions
accuracy on the original competition mathematics test set