Industrial design optimization on Unseen-category (test)
75FSRCOSMO-Agent
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| COSMO-AgentScale=8B2026.04 | 75 | 84 | 78 | 89 | 100 | 61.5 | 6.57 | |
| Gemini-3-FlashScale=-2026.04 | 57 | 60 | 57 | 60 | 60 | 69.77 | 9.44 | |
| Claude-Sonnet-4.5Scale=-2026.04 | 38 | 53 | 81 | 73 | 94 | 64.68 | 11.08 | |
| Intern-S1Scale=236B2026.04 | 35 | 53 | 79 | 63 | 99 | 56.88 | 7.56 | |
| Qwen3-30BScale=30B2026.04 | 28 | 45 | 81 | 71 | 100 | 58.7 | 8.41 | |
| Qwen3-NextScale=80B2026.04 | 24 | 43 | 80 | 57 | 100 | 56.9 | 8.71 | |
| Intern-S1-miniScale=8B2026.04 | 20 | 27 | 41 | 31 | 49 | 31.11 | 5.81 | |
| Llama-4-ScoutScale=17B2026.04 | 19 | 33 | 43 | 37 | 62 | 25.2 | 3.44 |