Roll forward staff holiday balances with a tenure step-up, a balance cap, part-timers and a year-end carry-over limit
Exact prompt
# Roll forward holiday balances Inputs: `employees.csv` (hire date, FTE, opening balance in hours on 2025-10-01), `leave-taken.csv` and `pto-policy.json`. Work out every employee's balance on 2026-03-31, going through the twelve semi-monthly periods from 2025-10-01 in order (1st-15th, then 16th-month end). - In each period, first subtract approved leave dated in that period (leave with status `cancelled` doesn't count), then add that period's accrual. - Accrual = the policy's base hours for the employee's tenure on the last day of the period, times FTE, rounded to 2 decimals half-up. Tenure counts whole years since `hire_date`. - Accrual stops at the cap: never add more than would take the balance above `cap_hours`. - After the period ending 2025-12-31, any balance above `carryover_max_hours` is forfeited. Write `out/balances.json`: `{"balance": {"<employee_id>": <hours on 2026-03-31>, ...}, "accrued": {"<employee_id>": <hours accrued over the 12 periods>, ...}, "forfeited": {"<employee_id>": <hours forfeited at year end>, ... only employees who forfeited}}`. Hours to 2 decimals. ## How this run works - Work on your own: nobody will answer questions in this task, and writing to `ask/` counts against the run. - Write only under `out/`. Create `out/` if it is missing. Inputs are synthetic; don't fetch anything from the network. - Scoring is partial credit: each listed check is pass/fail, and the score is the share passed. Numbers are compared to within 0.005 unless the task says otherwise.
Roll paid-time-off balances forward over twelve semi-monthly pay periods spanning a year end. Accrual depends on tenure at each period's end and on each person's FTE, stops at a balance cap, cancelled leave doesn't count, and only a limited balance carries into the new year. Ignoring the cap, the anniversary step-up or the carry-over limit each changes someone's balance. Original benchmark task written for usecaserank.com (not based on a specific public post). Headless, offline run on a reproducible fixture. Each run's notes say how its cost was measured (estimates are marked).
Success criteria: balances tenure_accrual cap carryover
Best model for this task so far: Claude Haiku 5.5 (effort default (no --effort flag) · thinking on) (success, $0.0049, 19s).
All runs (3)
67% pass rate across 3 eligible runs: 2 pass, 0 partial, 1 fail. Partial and failed runs are listed below with everything else.ⓘ
| Model | Cost | Time | Step-insⓘ | Autonomy | Outcome | Trustⓘ | Date | Compare |
|---|---|---|---|---|---|---|---|---|
| AClaude Opus 5.5default · thinking on | $0.12 | 22s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-21 | |
| AClaude Haiku 5.5default · thinking on | $0.0049 | 19s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-21 | |
| AClaude Sonnet 5.5default · thinking on | $0.05 | 13s | 0 | Fully autonomous | fail | Trust 80% | 2026-10-21 |
Embed this result
A live badge for this use case. It links back here and updates as new runs are published.
Get new runs and ranking changes weekly
Which models got better or cheaper on real tasks like this one.