PRIVATE TRIAL DELIVERABLE
No paired campaign has been run. Column B stays empty.
Reference Trial
Same declared model, same declared host, same ten cases. Prompt A against Prompt B. Twenty executions. Three critical cases replayed. Empty cell = not measured. Not a ranking.
| A | B | |
|---|---|---|
| Declared model | — | — |
| Declared host | — | — |
| Executions present | 0 / 20 | |
| corrigé / kept / cassé | 0 / 0 / 0 | |
Ten paired cases
| id | seed | A | B | verdict |
|---|---|---|---|---|
rt-01 | — | — | — | not_measured |
rt-02 | — | — | — | not_measured |
rt-03 | — | — | — | not_measured |
rt-04 | — | — | — | not_measured |
rt-05 | — | — | — | not_measured |
rt-06 | — | — | — | not_measured |
rt-07 | — | — | — | not_measured |
rt-08 | — | — | — | not_measured |
rt-09 | — | — | — | not_measured |
rt-10 | — | — | — | not_measured |
Critical rerun (3 cases): —
Exterior OLY32 debt
A complete exterior OLY32 is a signed native card from a non-ScoreIA host, at least 20 actions, five seals tracked, and a replay with at least two seals lit. The three current public oly-32 cards are technical or internal. Empty cell = not measured.
Complete exterior cards: 0 / 3
| card | model | host | actions | seals | missing |
|---|---|---|---|---|---|
oly-1b1343626c3d4b4a8910f5d2ea0bc36e | gpt-5-codex | codex-agent | 1 | 1/5 | tens_of_actions, replay_depth |
oly-bb01b7f6fe4a407d9d2a74025c8ad81f | scoreia-zero-action-conformance | scoreia-lab | 0 | 0/1 | exterior_agent, tens_of_actions, five_seals_tracked, replay_depth |
oly-ce9936df33c740e8b18ad2c7f5ad2f64 | scoreia-internal-proof-smoke | scoreia-lab | 0 | 0/5 | exterior_agent, tens_of_actions, replay_depth |