Skip to content

Benchmark

U-MATH integral_calc accuracy

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

Models measured
30
Scores recorded
90
Sources
1
IQ weight
2.00

Score distribution

32.3%
Mean
27.3%
Median
22.9%
Std dev
9.6%
10th percentile
67.3%
90th percentile
1.3%
Min
1.3%90.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.00 for chat models, measured over 30 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51Gemini 1.5 Prointegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH24.1%
52Qwen2.5 Math 7B Instructintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH24%
53Qwen2-VL (72B) Instructintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH23.6%
54Llama 3.1 8B Instructintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH22.7%
55Qwen2.5 Math 7B Instructintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH22.1%
56Qwen2-VL (72B) Instructintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH22%
57Llama 3.1 8B Instructintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH21.6%
58Nvidia Llama 3.1 Nemotron 70B Instructintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH21.3%
59Llama 3.3 70B Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH20.7%
60Nvidia Llama 3.1 Nemotron 70B Instructintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH20.7%
61Llama 3.1 8B Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH19%
62Nvidia Llama 3.1 Nemotron 70B Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH19%
63Qwen2.5 7B Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH19%
64GPT-4o miniintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH17.2%
65Qwen2 VL 7b Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH17.2%
66Qwen2.5 72B Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH17.2%
67Qwen2.5 Math 7B Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH17.2%
68GPT-4o miniintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH16.8%
69GPT-4o miniintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH16.7%
70Llama 3.1 70B Instructintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH16%
71Claude 3.5 Sonnetintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH15.5%
72Mistral Large Instruct 2411integral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH15.5%
73Llama 3.1 70B Instructintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH14.4%
74LFM-7bintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH13.8%
75Claude 3.5 Sonnetintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH13%
76Qwen2 VL 7b Instructintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH13%
77Claude 3.5 Sonnetintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH12%
78Llama 3.2 90B Vision Instructintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH11.3%
79Qwen2 VL 7b Instructintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH11.3%
80Llama 3.1 70B Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH10.3%
81Llama 3.2 90B Vision Instructintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH9.6%
82LFM-7bintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH6.7%
83Llama 3.2 90B Vision Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH5.2%
84Pixtral 12Bintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH4.7%
85Pixtral 12Bintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH4.3%
86LFM-7bintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH4%
87Llama 3.2 11B Vision Instructintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH3.4%
88Pixtral 12Bintegral_calc, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH3.4%
89Llama 3.2 11B Vision Instructintegral_calc, all, gpt-4o-2024-08-06· U-MATH and mu-MATH1.9%
90Llama 3.2 11B Vision Instructintegral_calc, text, gpt-4o-2024-08-06· U-MATH and mu-MATH1.3%

Showing 51–90 of 90 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)