Task-Efficient Routing on Science Curated Task Benchmark 1.0 (test)
0.0054Average CostForce Weak
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Force WeakStrategy=Force Weak2026.01 | 0.0054 | 96 | 90.2 | |
| CASTERStrategy=CASTER2026.01 | 0.0831 | 37.9 | 95.3 | |
| Force StrongStrategy=Force Strong2026.01 | 0.1339 | — | 95.2 |