# Label prior

baseline. A reference point, not a model.

| Board | Decision Score (95% range) | Rank | Accuracy | Calibration gap (ECE, pts) | Hand-off at 95% | $ per 1k decisions | Speed p95 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| [noul (PubMedQA)](https://jevals.com/noul/) | 0.0 (0.0 to 0.0) | rank 8 of 8 | 62.0% | 0.0 | — | — | — |
| [choice (Banking77)](https://jevals.com/choice/) | 0.0 (0.0 to 0.0) | rank 8 of 8 | 1.3% | 0.0 | — | — | — |
| [score (HelpSteer2 helpfulness)](https://jevals.com/score/) | 0.0 (0.0 to 0.0) | tied for rank 1 of 8 | 41.7% | 0.0 | — | — | — |

Decision Score: 100 = perfect, 0 = guessing the label base rates, below 0 = worse than that. Ranks: 1 + the number of rows significantly better (paired item bootstrap, 95%); rows that cannot be told apart share a rank. Hand-off at 95%: the share of decisions a model can take alone, at its own confidence threshold, while staying at least 95% right.

Setup: model id —; served as —; adapter —; reasoning —; temperature —; price as of —.

Source: https://jevals.com/models/label-prior/ · release 2026-09-18, suite 0.1.0 · data (CC-BY-4.0): https://jevals.com/data/releases/2026-09-18/board.json · cite as: Jevals (jevals.com), release 2026-09-18, suite 0.1.0. CC-BY-4.0.