Hierarchical Task Decomposition on MagicEval-Plan Context Inheritance 3
97.6Step ScoreMagicAgent-32B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MagicAgent-32BScale=MagicAgent Series2026.02 | 97.6 | 94.3 | 92.2 | |
| MagicAgent-30B-A3BScale=MagicAgent Series2026.02 | 97.6 | 94.6 | 90.2 | |
| GLM-4.7Scale=Ultra-Scale Models2026.02 | 94.6 | 96.6 | 88.2 | |
| DeepSeek-V3.1-nothinkScale=Ultra-Scale Models2026.02 | 89.5 | 96.3 | 89.9 | |
| Kimi-K2-InstructScale=Ultra-Scale Models2026.02 | 88.7 | 91.7 | 88.3 | |
| Qwen3-MAXScale=Ultra-Scale Models2026.02 | 88.3 | 92 | 90.7 | |
| Qwen3-235B-A22B-InstructScale=Ultra-Scale Models2026.02 | 82.1 | 90.8 | 88 | |
| ERNIE-4.5-21B-A3B-PTScale=Large-Scale Models2026.02 | 76.3 | 91.3 | 87.9 | |
| Qwen3-32B-nothinkScale=Large-Scale Models2026.02 | 67.7 | 90.7 | 87.5 | |
| Olmo-3.1-32B-InstructScale=Large-Scale Models2026.02 | 61.9 | 90.5 | 79.9 | |
| GLM-4.7-FlashScale=Large-Scale Models2026.02 | 56.4 | 85.9 | 73.1 | |
| Llama3.3-70B-InstructScale=Large-Scale Models2026.02 | 38.5 | 90.3 | 85.5 | |
| GPT-5.2Scale=Ultra-Scale Models2026.02 | 33.5 | 89.7 | 91.8 | |
| Qwen3-30B-A3B-InstructScale=Large-Scale Models2026.02 | 33.5 | 90.3 | 78.7 |