GSM8K
Grade School Math 8K · 8,500 linguistically diverse grade-school math word problems that require multi-step reasoning to solve.
The Frontier
Best score over time · one chart, every benchmark
Full rankings
43 models tested · sorted by score
| # | Model | Score |
|---|---|---|
| 1 | 92.0 | |
| 2 | 91.3 | |
| 3 | 91.3 | |
| 4 | 91.1 | |
| 5 | 90.0 | |
| 6 | 88.7 | |
| 7 | 86.7 | |
| 8 | 86.7 | |
| 9 | 86.7 | |
| 10 | 84.9 | |
| 11 | 84.9 | |
| 12 | 84.2 | |
| 13 | 82.4 | |
| 14 | 82.4 | |
| 15 | 75.7 | |
| 16 | 74.4 | |
| 17 | 74.4 | |
| 18 | U Stable Beluga 2 | 69.6 |
| 19 | 67.1 | |
| 20 | 65.8 | |
| 21 | U internlm-20b | 62.9 |
| 22 | 61.3 | |
| 23 | 57.8 | |
| 24 | U StarCoder 2 15B | 57.7 |
| 25 | 54.4 | |
| 26 | 54.4 | |
| 27 | 53.8 | |
| 28 | U Nemotron-4 15B | 46.0 |
| 29 | U Yi 6B | 44.9 |
| 30 | U INTELLECT-1 | 38.6 |
| 31 | 37.7 | |
| 32 | 36.9 | |
| 33 | 35.4 | |
| 34 | 34.5 | |
| 35 | U MPT-30B | 34.4 |
| 36 | U vicuna-13b-v1.1 | 28.1 |
| 37 | U Baichuan 2-7B | 24.6 |
| 38 | 21.3 | |
| 39 | 21.2 | |
| 40 | 20.6 | |
| 41 | 17.7 | |
| 42 | U Baichuan1-7B | 9.2 |
| 43 | 4.4 |
Score distribution
Where models cluster
Correlated benchmarks
Pearson r · original research
Benchmarks that track with GSM8K
Pearson correlation across models scored on both benchmarks. Closer to 1 = strongly predictive.
Frequently asked
About GSM8K
What does GSM8K measure?
Grade School Math 8K · 8,500 linguistically diverse grade-school math word problems that require multi-step reasoning to solve. 43 AI models have been tested on it. Scores range from 4.4 to 92.0 out of 100.
Which model leads on GSM8K?
GPT-4 (older v0314) from OpenAI leads GSM8K with a score of 92.0. The median score across 43 tested models is 61.3.
Is GSM8K saturated?
No · the top score is 92.0 out of 100 (92%). There is still meaningful room for improvement on GSM8K.
Does GSM8K predict performance on other benchmarks?
Yes · GSM8K scores correlate 0.89 with MMLU across 40 shared models. Models that do well on GSM8K tend to do well on MMLU.
How often is GSM8K data refreshed?
BenchGecko pulls updates daily. New model scores on GSM8K appear as soon as they are published by Epoch AI or the model provider.
- Category
- Math
- Max score
- 100
- Models
- 43
- Updated
- 2025-04-15
Top on GSM8K
GPT-4 (older v0314) · 92.0GPT-4o-mini · 91.3GPT-4o-mini (2024-07-18) · 91.3Qwen2.5 Coder 32B Instruct · 91.1GPT-4 Turbo · 90.0More math benchmarks
Same category · related evaluations