Spatiotemporal reasoning on ST-Bench 2026
55.3Exact Match (EM)STAR
Evaluation Results
| Method | Links | |
|---|---|---|
| STARBackbone=Qwen3-8B, Query Eligibility=EM-eligible queries2026.05 | 55.3 | |
| LLM-only (CoT)Backbone=Qwen3-8B, Query Eligibility=EM-eligible queries2026.05 | 45 | |
| Graph-of-ThoughtBackbone=Qwen3-8B, Query Eligibility=EM-eligible queries2026.05 | 42.5 | |
| ReflexionBackbone=Qwen3-8B, Query Eligibility=EM-eligible queries2026.05 | 40 | |
| Tree-of-ThoughtBackbone=Qwen3-8B, Query Eligibility=EM-eligible queries2026.05 | 40 | |
| LLM-onlyBackbone=Qwen3-8B, Query Eligibility=EM-eligible queries2026.05 | 37.5 | |
| Function-callingBackbone=Qwen3-8B, Query Eligibility=EM-eligible queries2026.05 | 33.3 | |
| ReActBackbone=Qwen3-8B, Query Eligibility=EM-eligible queries2026.05 | 26.7 |