Agentic Task Performance on ACEBench Agent (test)
56.7Average ScoreCURATEEVO
Evaluation results
| Method | Links | |
|---|---|---|
| CURATEEVOModel=QWEN3-4B, Data Setting=Labeled Data2026.07 | 56.7 | |
| CURATEEVOModel=QWEN3-4B, Data Setting=Wild Data2026.07 | 55.8 | |
| MUA-RLModel=QWEN3-8B, Data Setting=Labeled Data2026.07 | 53.3 | |
| MUA-RLModel=QWEN3-8B, Data Setting=Wild Data2026.07 | 50.7 | |
| EnvScalerModel=QWEN3-8B, Data Setting=Labeled Data2026.07 | 50.4 | |
| EnvScalerModel=QWEN3-8B, Data Setting=Wild Data2026.07 | 48.8 | |
| RODSModel=QWEN3-4B, Data Setting=Labeled Data2026.07 | 48.3 | |
| AWMModel=QWEN3-8B, Data Setting=Labeled Data2026.07 | 47.8 | |
| RODSModel=QWEN3-4B, Data Setting=Wild Data2026.07 | 46.2 | |
| AWMModel=QWEN3-8B, Data Setting=Wild Data2026.07 | 45.9 | |
| FunReason-MTModel=QWEN3-4B, Data Setting=Labeled Data2026.07 | 45.7 | |
| FunReason-MTModel=QWEN3-4B, Data Setting=Wild Data2026.07 | 43.4 | |
| GRPO w/o. Data CurationModel=QWEN3-8B, Data Setting=Labeled Data2026.07 | 36.8 | |
| GRPO w/o. Data CurationModel=QWEN3-4B, Data Setting=Labeled Data2026.07 | 32.1 | |
| GRPO w/o. Data CurationModel=QWEN3-8B, Data Setting=Wild Data2026.07 | 30.9 | |
| GRPO w/o. Data CurationModel=QWEN3-4B, Data Setting=Wild Data2026.07 | 28.2 |