| Benchmark | Model | Rank | Results |
|---|---|---|---|
| code-generation-on-humaneval | Loopie-20B-A2B (Thinking) | – | Pass@1: 89.02 |
| code-generation-on-mbpp | Loopie-20B-A2B (Thinking) | – | Accuracy: 89.49 |
| mathematical-reasoning-on-aime-2025 | Loopie-20B-A2B (Thinking) | – | Accuracy: 83.75 |
| mathematical-reasoning-on-aime24 | Loopie-20B-A2B (Thinking) | – | Acc: 92.09 |
| mathematical-reasoning-on-olympiadbench | Loopie-20B-A2B (Thinking) | – | Accuracy: 80.5 |
| multi-task-language-understanding-on-mmlu-redux | Loopie-20B-A2B (Thinking) | – | Accuracy: 83.61 |