Interactive Decision Making on ScienceWorld (OOD)
9.9ScoreMTRouter
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MTRouterRouting Level=Multi-Turn Routers (turn-level)2026.04 | 9.9 | 16.3 | |
| RouterDCRouting Level=Single-Turn Routers (episode-level)2026.04 | 5.5 | 2.5 | |
| EmbedLLMRouting Level=Single-Turn Routers (episode-level)2026.04 | 5 | 3 | |
| GPT-5Routing Level=Single-Model Baselines2026.04 | 4.9 | 47.6 | |
| AvengersProRouting Level=Single-Turn Routers (episode-level)2026.04 | 2.4 | 4.1 | |
| Router-R1Routing Level=Multi-Turn Routers (turn-level)2026.04 | 2.1 | 21 | |
| GPT-OSS-120BRouting Level=Single-Model Baselines2026.04 | 1.1 | 4.2 | |
| MiniMax-M2Routing Level=Single-Model Baselines2026.04 | 0.9 | 10.9 | |
| Kimi-K2Routing Level=Single-Model Baselines2026.04 | 0.2 | 8.9 | |
| LLM RouterRouting Level=Multi-Turn Routers (turn-level)2026.04 | -0.4 | 28.3 | |
| Gemini-2.5-Flash-LiteRouting Level=Single-Model Baselines2026.04 | -2.1 | 1.5 | |
| DeepSeek-V3.2Routing Level=Single-Model Baselines2026.04 | -4.2 | 22.8 | |
| Random RouterRouting Level=Multi-Turn Routers (turn-level)2026.04 | -8.1 | 20.3 | |
| OpenRouterRouting Level=Multi-Turn Routers (turn-level)2026.04 | -26.9 | 15.5 |