Hypersonic Thermal Protection System design and analysis on HyTPS-Bench
88.7EESRAeroTherm-GPT
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| AeroTherm-GPTScaffolding Framework=Constraint-Closed-Loop Generation2026.04 | 88.7 | 84.7 | 92.4 | 87.1 | 84.3 | 87.9 | 96.8 | 93.7 | |
| CDG-SFT AgentScaffolding Framework=Constraint Dependency Graph, Training/Alignment Protocol=SFT2026.04 | 78.4 | 71.4 | 88.3 | 83.6 | 80.4 | 80.4 | 87.4 | 61.3 | |
| Flat-VER + SFT + DPOScaffolding Framework=Flat-VER, Training/Alignment Protocol=SFT + DPO2026.04 | 76.2 | 73.1 | 89.1 | 84.3 | 81.3 | 81.9 | 88.6 | 63.7 | |
| GPT-5 (high)Thinking Mode=with thinking2026.04 | 48 | 54 | 67.2 | 61.8 | 56.8 | 57.6 | 72.8 | 35.9 | |
| Claude-4.1-Opus (w/)Thinking Mode=with thinking2026.04 | 47 | 54 | 80.3 | 48.7 | 45.7 | 55.1 | 68.3 | 33.1 | |
| Flat-VERScaffolding Framework=Flat-VER2026.04 | 46.8 | 63.8 | 85.4 | 80.3 | 77.4 | 76.1 | 81.2 | 48.9 | |
| GPT-4o (RAG+Few Shot)Prompting Strategy=RAG+Few Shot2026.04 | 45.2 | 59.1 | 87.1 | 82.4 | 75.4 | 69.8 | 80.7 | 37.9 | |
| Claude-4.1-Opus (w/o)Thinking Mode=without thinking2026.04 | 45 | 55 | 55.1 | 74.1 | 69.1 | 59.7 | 70.5 | 33.7 | |
| Claude-4-Sonnet (w/)Thinking Mode=with thinking2026.04 | 45 | 53 | 76.8 | 42.1 | 40.1 | 51.4 | 65.7 | 31.2 | |
| Tool-AgentScaffolding Framework=Tool-Agent2026.04 | 44.1 | 61.2 | 84.7 | 78.8 | 76.2 | 74.3 | 78.3 | 46.1 | |
| Claude-4-Sonnet (w/o)Thinking Mode=without thinking2026.04 | 42 | 48 | 49.3 | 71.2 | 64.3 | 55 | 68.2 | 32.1 | |
| GLM-4.5 (w/)Thinking Mode=with thinking2026.04 | 42 | 54 | 42.1 | 36.4 | 34.1 | 41.7 | 61.5 | 29.8 | |
| Qwen3-235B-TThinking Mode=with thinking2026.04 | 40 | 48 | 56.2 | 49.7 | 46.3 | 48 | 63.2 | 28.7 | |
| GLM-4.5 (w/o)Thinking Mode=without thinking2026.04 | 38 | 50 | 45.2 | 39.7 | 37.3 | 42 | 60.3 | 29.2 | |
| GPT-4o+Refl.Scaffolding Framework=Reflection, Base Model=GPT-4o2026.04 | 37.6 | 53.7 | 83.1 | 77.4 | 75.1 | 72.9 | 75.8 | 43.7 | |
| DeepSeek-R1-0528Thinking Mode=with thinking2026.04 | 37 | 43 | 38.1 | 31.7 | 30.8 | 36.1 | 59.1 | 26.4 | |
| GPT-5 (min)Thinking Mode=without thinking2026.04 | 35 | 29 | 64.1 | 57.3 | 52.1 | 47.5 | 62.3 | 28.5 | |
| DeepSeek-V3-0324Thinking Mode=without thinking2026.04 | 32 | 36 | 57.1 | 50.4 | 45.2 | 44.1 | 58.9 | 26.8 | |
| C3.5 (Zero Shot)Prompting Strategy=Zero Shot2026.04 | 30.7 | 48.3 | 84.3 | 77.9 | 70.6 | 62.4 | 70.9 | 34.8 | |
| GPT-4o (Zero Shot)Prompting Strategy=Zero Shot2026.04 | 28.1 | 42.7 | 81.2 | 75.3 | 67.1 | 58.9 | 67.8 | 31.7 | |
| Qwen3-32B (w/o)Thinking Mode=without thinking2026.04 | 25 | 23 | 38.7 | 33.1 | 29.3 | 29.8 | 51.7 | 22.4 |