Skip to content

Benchmark

mu-MATH PPV

Math benchmarks measure how well models solve maths problems. Scores are percentages. Higher is better.

Models measured
17
Scores recorded
85
Sources
1
IQ weight
1.50

Score distribution

77.7%
Mean
81.2%
Median
11.5%
Std dev
62.8%
10th percentile
89.7%
90th percentile
35.5%
Min
35.5%94.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.50 for chat models, measured over 17 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 1.5 FlashGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH94.8%
2GPT-4o miniGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH92.2%
3Claude 3.5 SonnetGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH91.8%
4Gemini 1.5 ProGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH91.6%
5o1Gemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH91.1%
6GPT 4oGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH90.5%
7o1Llama-3.1-70B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH90.5%
8O1 MiniGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH90.5%
9DeepSeek R1Gemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH89.7%
10o1GPT-4o, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH88.9%
11o1mu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH88.7%
12Qwen2.5 Math 72B InstructGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH87.6%
13Claude 3.5 Sonnetmu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH87.3%
14Mistral Large Instruct 2411Gemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH87.3%
15Gemini 2.0 Flash ExperimentalGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct, thinking· U-MATH and mu-MATH87%
16DeepSeek R1GPT-4o, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH86.7%
17Qwen2.5 72B InstructGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH86.7%
18Claude 3.5 SonnetQwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH86.5%
19DeepSeek R1mu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH86.2%
20Gemini 1.5 Flashmu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH86.2%
21O1 Minimu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH85.8%
22O1 MiniLlama-3.1-70B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH85.7%
23QwQ 32B PreviewQwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH85.4%
24GPT-4o miniQwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH85.3%
25Gemini 1.5 Promu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH85.2%
26GPT 4omu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH85.1%
27GPT-4o minimu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH85.1%
28O1 MiniGPT-4o, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH84.8%
29Gemini 1.5 FlashQwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH84.7%
30GPT 4oQwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH84.6%
31o1Qwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH84.6%
32Claude 3.5 SonnetGPT-4o, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH84.4%
33DeepSeek R1Qwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH84.4%
34Gemini 1.5 ProQwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH84.2%
35Claude 3.5 SonnetLlama-3.1-70B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH82.9%
36Gemini 1.5 FlashLlama-3.1-70B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH82.3%
37GPT 4oGPT-4o, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH82.1%
38Gemini 1.5 ProLlama-3.1-70B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH82%
39Qwen2.5 7B InstructGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH82%
40DeepSeek R1Llama-3.1-70B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH81.8%
41O1 MiniQwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH81.6%
42GPT-4o miniGPT-4o, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH81.4%
43GPT 4oLlama-3.1-70B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH81.2%
44Gemini 1.5 ProGPT-4o, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH80.5%
45QwQ 32B PreviewGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH80.4%
46Mistral Large Instruct 2411mu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH79.7%
47Gemini 1.5 FlashGPT-4o, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH79.4%
48Mistral Large Instruct 2411Qwen2.5-72B-Instruct, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH79.1%
49QwQ 32B Previewmu_math, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH78.7%
50Llama 3.1 70B InstructGemini-1.5-Pro, PPV, Qwen/Qwen2.5-72B-Instruct· U-MATH and mu-MATH78.4%

Showing 1–50 of 85 results.

Data credits: Toloka, U-MATH and mu-MATH public leaderboard. MIT Space; generated solution outputs retain underlying model terms and are not ingested. (licence)