| Benchmark | Model | Rank | Results |
|---|---|---|---|
| code-generation-on-livecodebench-v6 | MetaAgent-X RL (Qwen3 8B) | – | Pass@1: 41.0 |
| mathematical-reasoning-on-aime-2025 | MetaAgent-X RL (Qwen3 8B) | – | Accuracy: 33.33 |
| mathematical-reasoning-on-olympiadbench | MetaAgent-X RL (Qwen3 8B) | – | Accuracy: 61.0 |