# Mercury 2.5 (Inception)

LLM, Inception, no reasoning. Verbalized probabilities through the Jevals adapter, via OpenRouter · Inception.

| 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/) | 55.7 (44.8 to 65.4) | tied for rank 3 of 8 | 87.1% | 4.7 | 19% | $0.023 | 1.0 s |
| [choice (Banking77)](https://jevals.com/choice/) | 54.0 (47.8 to 60.6) | rank 7 of 8 | 70.3% | 10.4 | — | $0.036 | 1.5 s |
| [score (HelpSteer2 helpfulness)](https://jevals.com/score/) | −5.5 (−16.2 to 4.0) | tied for rank 3 of 8 | 41.7% | 31.5 | — | $0.034 | 1.2 s |

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 inception/mercury-2.5; served as inception/mercury-2.5; adapter jevals-llm 0383a0e3e592; reasoning none; temperature provider default; price as of 2026-09-18.

Compare: [Jev vs Mercury 2.5](https://jevals.com/compare/jev-vs-mercury-2.5/)

Source: https://jevals.com/models/mercury-2.5/ · 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.