Benchmark
U-MATH algebra 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
- 2.19
Score distribution
- 59.7%
- Mean
- 72.2%
- Median
- 31%
- Std dev
- 10%
- 10th percentile
- 92%
- 90th percentile
- 0%
- Min
This benchmark has a weight of 2.19 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
| # | Model | Score |
|---|---|---|
| 1 | o3 Minialgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 99.3% |
| 2 | o1algebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 97.3% |
| 3 | O1 Minialgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 97.3% |
| 4 | DeepSeek R1algebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 96.7% |
| 5 | DeepSeek V3algebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 96% |
| 6 | Gemini 2.0 Flash Experimentalalgebra, text, gpt-4o-2024-08-06, thinking· U-MATH and mu-MATH | 95.3% |
| 7 | QwQ 32B Previewalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 95.3% |
| 8 | Qwen2.5 Math 72B Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 94.7% |
| 9 | QVQ 72B Previewalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 94% |
| 10 | Qwen2.5 32B Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 92% |
| 11 | Gemini 1.5 Proalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 91.3% |
| 12 | Gemini 1.5 Flashalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 90.7% |
| 13 | Qwen2.5 72B Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 90.7% |
| 14 | GPT 4oalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 90% |
| 15 | Gemini 2.0 Flash Experimentalalgebra, all, gpt-4o-2024-08-06, thinking· U-MATH and mu-MATH | 89.4% |
| 16 | o1algebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 89.4% |
| 17 | Phi 4algebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 89.3% |
| 18 | O1 Minialgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 87.8% |
| 19 | GPT-4o minialgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 87.3% |
| 20 | Gemini 1.5 Proalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 86.1% |
| 21 | Qwen2.5 7B Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 86% |
| 22 | Mistral Large Instruct 2411algebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 85.3% |
| 23 | Qwen2.5 Math 7B Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 84.7% |
| 24 | o3 Minialgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 84.4% |
| 25 | Nvidia Llama 3.1 Nemotron 70B Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 84% |
| 26 | QVQ 72B Previewalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 83.9% |
| 27 | DeepSeek R1algebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 83.3% |
| 28 | Gemini 1.5 Flashalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 83.3% |
| 29 | Llama 3.3 70B Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 83.3% |
| 30 | Llama 3.2 90B Vision Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 82% |
| 31 | DeepSeek V3algebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 81.7% |
| 32 | Phi 4algebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 81.7% |
| 33 | GPT 4oalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 80.6% |
| 34 | Qwen2-VL (72B) Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 80% |
| 35 | Qwen2.5 Math 72B Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 80% |
| 36 | QwQ 32B Previewalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 80% |
| 37 | Qwen2.5 32B Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 78.9% |
| 38 | Llama 3.1 70B Instructalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 78.7% |
| 39 | Qwen2.5 72B Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 78.3% |
| 40 | Claude 3.5 Sonnetalgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 75.3% |
| 41 | GPT-4o minialgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 75% |
| 42 | Qwen2.5 7B Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 75% |
| 43 | Nvidia Llama 3.1 Nemotron 70B Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 73.9% |
| 44 | Mistral Large Instruct 2411algebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 73.3% |
| 45 | Llama 3.2 90B Vision Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 72.2% |
| 46 | Qwen2.5 Math 7B Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 71.7% |
| 47 | Qwen2-VL (72B) Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 71.1% |
| 48 | Llama 3.3 70B Instructalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 70.6% |
| 49 | Claude 3.5 Sonnetalgebra, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 67.8% |
| 50 | LFM-7balgebra, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 66.7% |
Showing 1–50 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)