# Test-Time Scaling Selection on RouterBench Quality-Priority Source: [Test-Time Scaling Selection on RouterBench Quality-Priority]() Collection index: [llms.txt]() Data updated: 2026-06-01T08:34:35.500Z Evaluation of model performance focusing on selecting the appropriate scaling configuration during test time, using the Quality-Priority variant of the RouterBench dataset. Metrics include Reward, Accuracy, and Cost. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Test-Time Scaling Selection | | Dataset (including any stated split) | RouterBench Quality-Priority | | Metric | Reward | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 16 of 16 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Reward](): Higher is better; unit: SCALAR. - [Accuracy](): Higher is better; unit: PERCENT. - [Cost](): Lower is better; unit: SCALAR. ## Recorded results | Method | Reward | Result date | Report paper | | --- | --- | --- | --- | | Oracle | 0.7924 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | Oracle | 0.7426 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | UNISCALE | 0.645 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | UNISCALE | 0.6306 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | TS | 0.6243 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | Greedy | 0.6184 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | Random | 0.6175 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | MLP | 0.6055 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | NeuralUCB | 0.5929 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | TS | 0.5901 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | k-NN | 0.5807 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | Greedy | 0.578 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | MLP | 0.5752 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | Random | 0.5726 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | NeuralUCB | 0.558 | Not recorded | 6a1d11e0dbd7150506f83dd0 | | k-NN | 0.5146 | Not recorded | 6a1d11e0dbd7150506f83dd0 | ### Oracle Reward: 0.7924. Result ID: 6a1d42ecfbc373592c167318. [Reporting paper]() Recorded evaluation configuration: {"Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### Oracle Reward: 0.7426. Result ID: 6a1d42ecfbc373592c167308. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B","Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### UNISCALE Reward: 0.645. Result ID: 6a1d42ecfbc373592c167307. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B","Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### UNISCALE Reward: 0.6306. Result ID: 6a1d42ecfbc373592c167317. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### TS Reward: 0.6243. Result ID: 6a1d42ecfbc373592c167316. [Reporting paper]() Recorded evaluation configuration: {"Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### Greedy Reward: 0.6184. Result ID: 6a1d42ecfbc373592c167302. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B","Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### Random Reward: 0.6175. Result ID: 6a1d42ecfbc373592c167311. [Reporting paper]() Recorded evaluation configuration: {"Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### MLP Reward: 0.6055. Result ID: 6a1d42ecfbc373592c167313. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### NeuralUCB Reward: 0.5929. Result ID: 6a1d42ecfbc373592c167315. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### TS Reward: 0.5901. Result ID: 6a1d42ecfbc373592c167306. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B","Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### k-NN Reward: 0.5807. Result ID: 6a1d42ecfbc373592c167314. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### Greedy Reward: 0.578. Result ID: 6a1d42ecfbc373592c167312. [Reporting paper]() Recorded evaluation configuration: {"Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### MLP Reward: 0.5752. Result ID: 6a1d42ecfbc373592c167303. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B","Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### Random Reward: 0.5726. Result ID: 6a1d42ecfbc373592c167301. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B","Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### NeuralUCB Reward: 0.558. Result ID: 6a1d42ecfbc373592c167305. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B","Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. ### k-NN Reward: 0.5146. Result ID: 6a1d42ecfbc373592c167304. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B","Number of seeds":"5","Iterations":"160","Warm-up phase":"50-step"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.