Reasoning on HLE OOD
38.6AccuracyMTRouter
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MTRouterRouting Level=Multi-Turn Routers (turn-level)2026.04 | 38.6 | 31.2 | |
| LLM RouterRouting Level=Multi-Turn Routers (turn-level)2026.04 | 36 | 35.6 | |
| Router-R1Routing Level=Multi-Turn Routers (turn-level)2026.04 | 35.1 | 60.7 | |
| GPT-5Routing Level=Single-Model Baselines2026.04 | 34.8 | 65.3 | |
| OpenRouterRouting Level=Multi-Turn Routers (turn-level)2026.04 | 34 | 154.3 | |
| EmbedLLMRouting Level=Single-Turn Routers (episode-level)2026.04 | 33.6 | 56.8 | |
| AvengersProRouting Level=Single-Turn Routers (episode-level)2026.04 | 30.6 | 33.3 | |
| DeepSeek-V3.2Routing Level=Single-Model Baselines2026.04 | 28.7 | 22.2 | |
| Random RouterRouting Level=Multi-Turn Routers (turn-level)2026.04 | 23.8 | 14.7 | |
| Kimi-K2Routing Level=Single-Model Baselines2026.04 | 20.1 | 9.8 | |
| RouterDCRouting Level=Single-Turn Routers (episode-level)2026.04 | 17.9 | 13.5 | |
| GPT-OSS-120BRouting Level=Single-Model Baselines2026.04 | 11.4 | 2.1 | |
| MiniMax-M2Routing Level=Single-Model Baselines2026.04 | 8.9 | 9.8 | |
| Gemini-2.5-Flash-LiteRouting Level=Single-Model Baselines2026.04 | 8.4 | 2.1 |