| Benchmark | Model | Rank | Results |
|---|---|---|---|
| code-generation-on-humaneval | Prism (LLaDA-8B-Instruct, K=8) | #64 | Pass@1: 79.27 |
| arithmetic-reasoning-on-gsm8k | Prism (LLaDA-8B-Instruct, K=8) | #114 | Accuracy: 85.3 |
| code-generation-on-mbpp | Prism (LLaDA-8B-Instruct, K=8) | #139 | Accuracy: 38.2 |