# Sphragis certificate — ✅ PASS

> For every axis, the candidate was demonstrated non-inferior to the reference: the one-sided 95% upper confidence bound on the accuracy regression is below the margin of 3.0%.

- **Reference:** `ref` @ `http://127.0.0.1:8080/v1`
- **Candidate:** `cand` @ `http://127.0.0.1:8081/v1`
- **Pack:** /root/pack_v2.jsonl (2800 items, sha256 `2de27099bbb15bab…`)
- **Params:** margin 3.0%, alpha 0.05, n_floor 30, seed 0
- **Generated:** 2026-07-27T03:53:12.900485+00:00 by sphragis 0.1.0

| Axis | n | ref acc | cand acc | regression d | 95% CI | d upper bound | p (regr., Holm) | verdict |
|---|---|---|---|---|---|---|---|---|
| arithmetic | 1400 | 0.932 | 0.930 | +0.002 | [-0.001, +0.006] | +0.005 | 0.75 | ✅ PASS (non_inferior_within_margin) |
| instruction | 600 | 0.757 | 0.757 | +0.000 | [+0.000, +0.000] | +0.000 | 1 | ✅ PASS (non_inferior_within_margin) |
| knowledge | 400 | 0.968 | 0.973 | -0.005 | [-0.025, +0.015] | +0.010 | 1 | ✅ PASS (non_inferior_within_margin) |
| reasoning | 400 | 0.927 | 0.932 | -0.005 | [-0.013, +0.000] | +0.000 | 1 | ✅ PASS (non_inferior_within_margin) |