Task Routing on Data
0.0136Cost ($)Force Weak
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Force WeakModel=qwen2026.01 | 0.0136 | 70.2 | 70.4 | |
| CASTERModel=qwen2026.01 | 0.0344 | 24.6 | 73.6 | |
| Force WeakModel=openai2026.01 | 0.0452 | 96.1 | 73.1 | |
| Force StrongModel=qwen2026.01 | 0.0456 | — | 73.5 | |
| CASTERModel=deepseek2026.01 | 0.1332 | 39.3 | 78.1 | |
| Force StrongModel=deepseek2026.01 | 0.2193 | — | 73.4 | |
| Force WeakModel=deepseek2026.01 | 0.2268 | -3.4 | 69.2 | |
| CASTERModel=openai2026.01 | 0.4533 | 60.6 | 75.4 | |
| Force WeakModel=claude2026.01 | 0.4641 | 85.6 | 80.7 | |
| Force WeakModel=gemini2026.01 | 0.4665 | 74.2 | 52.1 | |
| CASTERModel=claude2026.01 | 0.9186 | 71.5 | 84.6 | |
| CASTERModel=gemini2026.01 | 0.9466 | 47.6 | 53.6 | |
| Force StrongModel=openai2026.01 | 1.1514 | — | 75.5 | |
| Force StrongModel=gemini2026.01 | 1.8052 | — | 53.6 | |
| Force StrongModel=claude2026.01 | 3.2237 | — | 83.3 |