ai4i-maintenance-leakage-imbalance-cost

Task gallery / Machine learning / Version 1

Reference solution (oracle): Pass
No-op resistance check: Pass
QA verdict: Not recorded
Aggregate eligibility: Eligible version

The task

The reliability team runs a condition-monitoring model that flags which production units are likely to fail so a unit can be pulled for inspection before it takes the line down. The model shipped last quarter and the team escalated when the unplanned downtime it was meant to prevent barely moved, even though the launch review had shown near-perfect detection. You have inherited the pipeline that produced that review…

Excerpt only; not a complete executable task specification. Task ID task_d525db2f015c4333 · version 1. Selection status: not fully recorded.

Results by model and effort

This exact task version only. Every rate is passes / valid runs with a Wilson 95% interval; n≤2 remains counts only. Macro aggregation does not apply within a single task. Baselines are summarized in the quality gates, not ranked. A passed no-op resistance check means that doing nothing did not solve the task.

Current evidence

Model / effort	Verifier result [95% Wilson]	Diagnostic labels / valid	HARNESS_ERROR · excluded	Outcome classes
Claude Opus 5 · default (unverified)	0/7 · 0.0% [0.0%, 35.4%]	7 / 7	0	GOOD_FAILURE: 6 BAD_FAILURE: 1 GOOD_SUCCESS: 0 BAD_SUCCESS: 0

Legacy evidence

Model / effort	Verifier result [95% Wilson]	Diagnostic labels / valid	HARNESS_ERROR · excluded	Outcome classes

Difficulty, with uncertainty

Selected-best difficulty: 100.0% [64.6%, 100.0%]. This is one minus the best measured current-row pass rate among rows with at least five valid runs.

The maximum is selected from the observed rows, so the estimate is optimistic. Its Wilson interval is not adjusted for model selection. “Hard” requires qualifying GOOD_FAILURE evidence; the threshold flag additionally requires the interval and recorded quality gates to meet the documented rule. This is not independent validation. Threshold flag: yes.

Every recorded run

13 recorded trials, including controls. Diagnostic labels do not replace verifier reward. Labels and quality checks are source declarations, not independently attested here. Internal evidence remains private; a missing public URL is shown explicitly. Records with invalid or absent rewards are not scored.

Run / model	Date / source	Reward / eligibility	Diagnostic class / source	Resources	Evidence
`trial_37772646fb094bfb` oracle@@	Oct 11, 2026 console	1 · valid	No diagnostic label derived	Tokens: — Turns: — Cost: —	Not publicly available
`trial_b5d61e8794cf47e2` nop@@	Oct 11, 2026 console	0 · valid	No diagnostic label derived	Tokens: — Turns: — Cost: —	Not publicly available
`trial_9b5f9aa64ef94e9e` Claude Opus 5 · default (unverified)	Oct 11, 2026 console	0 · valid	BAD_FAILURE frozen-classifier	Tokens: 26,904 Turns: 26 Cost: $1.6386	Not publicly available
`trial_f33d276283ec4c8c` Claude Opus 5 · default (unverified)	Oct 11, 2026 console	0 · valid	GOOD_FAILURE frozen-classifier	Tokens: 41,018 Turns: 27 Cost: $2.7917	Not publicly available
`trial_20ea88e04a034662` Claude Opus 5 · default (unverified)	Oct 11, 2026 console	0 · valid	GOOD_FAILURE frozen-classifier	Tokens: 35,119 Turns: 24 Cost: $1.9878	Not publicly available
`trial_16f1aa8d0e12492e` Claude Opus 5 · default (unverified)	Oct 11, 2026 console	0 · valid	GOOD_FAILURE frozen-classifier	Tokens: 61,962 Turns: 45 Cost: $4.3012	Not publicly available
`trial_a7644d05b6ca40ee` Claude Opus 5 · default (unverified)	Oct 11, 2026 console	0 · valid	GOOD_FAILURE frozen-classifier	Tokens: 28,876 Turns: 24 Cost: $1.6779	Not publicly available
`trial_ac6169ecc1644334` Claude Opus 5 · default (unverified)	Oct 12, 2026 console	0 · valid	GOOD_FAILURE frozen-classifier	Tokens: 37,724 Turns: 27 Cost: $2.2469	Not publicly available
`trial_4715807b475048d1` Claude Opus 5 · default (unverified)	Oct 12, 2026 console	0 · valid	GOOD_FAILURE frozen-classifier	Tokens: 44,918 Turns: 31 Cost: $2.6807	Not publicly available
`trial_46093cf98c184304` Claude Opus 5 · default (unverified)	Oct 12, 2026 console	No reward · excludedno_reward	No diagnostic label No diagnostic source	Tokens: — Turns: — Cost: —	Not publicly available
`trial_529abaa5ac804a09` Claude Opus 5 · default (unverified)	Oct 12, 2026 console	No reward · excludedno_reward	No diagnostic label No diagnostic source	Tokens: — Turns: — Cost: —	Not publicly available
`trial_81fbcf056ed44b8c` Claude Opus 5 · default (unverified)	Oct 12, 2026 console	No reward · excludedno_reward	No diagnostic label No diagnostic source	Tokens: — Turns: — Cost: —	Not publicly available
`trial_620d7e1323184c69` Claude Opus 5 · default (unverified)	Oct 12, 2026 console	No reward · excludedno_reward	No diagnostic label No diagnostic source	Tokens: — Turns: — Cost: —	Not publicly available