Benchmark
U-MATH sequences_series 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
- 1.63
Score distribution
- 44.8%
- Mean
- 48.7%
- Median
- 27.3%
- Std dev
- 0%
- 10th percentile
- 84.4%
- 90th percentile
- 0%
- Min
This benchmark has a weight of 1.63 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 Minisequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 94% |
| 2 | o3 Minisequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 93.5% |
| 3 | DeepSeek R1sequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 93.3% |
| 4 | o1sequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 93.3% |
| 5 | DeepSeek R1sequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 92.9% |
| 6 | o1sequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 92.9% |
| 7 | Gemini 2.0 Flash Experimentalsequences_series, text, gpt-4o-2024-08-06, thinking· U-MATH and mu-MATH | 90% |
| 8 | Gemini 2.0 Flash Experimentalsequences_series, all, gpt-4o-2024-08-06, thinking· U-MATH and mu-MATH | 88.3% |
| 9 | QwQ 32B Previewsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 85.3% |
| 10 | QwQ 32B Previewsequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 84.4% |
| 11 | O1 Minisequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 77.3% |
| 12 | O1 Minisequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 76.6% |
| 13 | DeepSeek R1sequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 75% |
| 14 | o1sequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 75% |
| 15 | o3 Minisequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 75% |
| 16 | Qwen2.5 Math 72B Instructsequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 75% |
| 17 | DeepSeek V3sequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 72.7% |
| 18 | DeepSeek V3sequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 72.1% |
| 19 | Qwen2.5 Math 72B Instructsequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 68.8% |
| 20 | Qwen2.5 Math 72B Instructsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 68.7% |
| 21 | QVQ 72B Previewsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 68% |
| 22 | QVQ 72B Previewsequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 66.2% |
| 23 | Gemini 1.5 Prosequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 63.3% |
| 24 | Gemini 1.5 Prosequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 63% |
| 25 | Gemini 1.5 Flashsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 60.7% |
| 26 | Gemini 1.5 Flashsequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 60.4% |
| 27 | Qwen2.5 32B Instructsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 58% |
| 28 | Qwen2.5 32B Instructsequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 56.5% |
| 29 | Mistral Large Instruct 2411sequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 56% |
| 30 | Mistral Large Instruct 2411sequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 55.2% |
| 31 | Qwen2.5 72B Instructsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 54.7% |
| 32 | Qwen2.5 72B Instructsequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 54.5% |
| 33 | Qwen2.5 Math 7B Instructsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 52% |
| 34 | Qwen2.5 Math 7B Instructsequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 51.9% |
| 35 | DeepSeek V3sequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 36 | Gemini 1.5 Flashsequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 37 | Gemini 1.5 Prosequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 38 | GPT-4o minisequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 39 | Llama 3.1 70B Instructsequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 40 | O1 Minisequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 41 | Phi 4sequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 42 | Qwen2.5 72B Instructsequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 43 | Qwen2.5 Math 7B Instructsequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 44 | QwQ 32B Previewsequences_series, visual, gpt-4o-2024-08-06· U-MATH and mu-MATH | 50% |
| 45 | Phi 4sequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 48.7% |
| 46 | GPT 4osequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 46.7% |
| 47 | Llama 3.3 70B Instructsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 46.7% |
| 48 | Llama 3.3 70B Instructsequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 46.1% |
| 49 | GPT 4osequences_series, all, gpt-4o-2024-08-06· U-MATH and mu-MATH | 45.5% |
| 50 | Nvidia Llama 3.1 Nemotron 70B Instructsequences_series, text, gpt-4o-2024-08-06· U-MATH and mu-MATH | 43.3% |
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)