Task Routing on Science
0.0276Cost ($)Force Weak
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Force WeakModel=qwen2026.01 | 0.0276 | 73.1 | 97.5 | |
| Force WeakModel=openai2026.01 | 0.055 | 95.7 | 87.5 | |
| CASTERModel=qwen2026.01 | 0.0695 | 32.3 | 97.6 | |
| Force StrongModel=qwen2026.01 | 0.1026 | — | 96.7 | |
| CASTERModel=deepseek2026.01 | 0.1429 | 13.3 | 97.5 | |
| Force StrongModel=deepseek2026.01 | 0.1648 | — | 84.6 | |
| Force WeakModel=deepseek2026.01 | 0.166 | -0.7 | 89.5 | |
| Force WeakModel=gemini2026.01 | 0.2311 | 80.3 | 95 | |
| Force WeakModel=claude2026.01 | 0.4987 | 69.8 | 94.8 | |
| CASTERModel=gemini2026.01 | 0.9259 | 21.1 | 95 | |
| CASTERModel=openai2026.01 | 1.0416 | 17.8 | 95.4 | |
| CASTERModel=claude2026.01 | 1.1596 | 29.8 | 95.8 | |
| Force StrongModel=gemini2026.01 | 1.1737 | — | 94.2 | |
| Force StrongModel=openai2026.01 | 1.2679 | — | 95.3 | |
| Force StrongModel=claude2026.01 | 1.653 | — | 96.7 |