Task Efficient Routing on Science scenario
0.0054CostForce Weak
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Force Weak2026.01 | 0.0054 | 96 | |
| Force WeakModel=qwen2026.01 | 0.0276 | 73.1 | |
| Force WeakModel=openai2026.01 | 0.055 | 95.7 | |
| CASTERModel=qwen2026.01 | 0.0695 | 32.3 | |
| CASTER2026.01 | 0.0831 | 37.9 | |
| Force StrongModel=qwen2026.01 | 0.1026 | — | |
| Force Strong2026.01 | 0.1339 | — | |
| CASTERModel=deepseek2026.01 | 0.1429 | 13.3 | |
| Force StrongModel=deepseek2026.01 | 0.1648 | — | |
| Force WeakModel=deepseek2026.01 | 0.166 | -0.7 | |
| Force WeakModel=gemini2026.01 | 0.2311 | 80.3 | |
| Force WeakModel=claude2026.01 | 0.4987 | 69.8 | |
| CASTERModel=gemini2026.01 | 0.9259 | 21.1 | |
| CASTERModel=openai2026.01 | 1.0416 | 17.8 | |
| CASTERModel=claude2026.01 | 1.1596 | 29.8 | |
| Force StrongModel=gemini2026.01 | 1.1737 | — | |
| Force StrongModel=openai2026.01 | 1.2679 | — | |
| Force StrongModel=claude2026.01 | 1.653 | — |