Skip to content

Cost a restaurant's menu dishes from recipes with trim yields and unit conversions, and build the week's supplier order in whole cases

Exact prompt

# Cost a restaurant's menu dishes from recipes with trim yields and unit conversions, and build the week's supplier order in whole cases Inputs: `ingredients.csv` (ingredient, case_size, case_unit, case_price, yield_pct, on_hand), `recipes.json` (each dish's menu_price and recipe lines of ingredient, qty, unit), `forecast.csv` (dish, portions for next week) and `order-rules.json` (safety_pct). All data is synthetic; amounts are in dollars. These rules are the kitchen's policy; where `chef-note.txt` disagrees, the rules win. Rules: - Units: 1 kg = 1000 g and 1 L = 1000 ml; each is a count. A recipe line's qty is the usable amount after trimming; case_size and on_hand are in the ingredient's case_unit. - Plate cost: a recipe line costs (qty converted to the case_unit) / (yield_pct / 100) × case_price / case_size. A dish's plate cost is the exact sum of its lines, rounded half-up to the cent once (don't round line by line). - Food cost: food_cost_pct = the dish's rounded plate cost / menu_price × 100, rounded half-up to one decimal place. - Ordering: an ingredient's purchase need for the week is the sum over dishes of portions × qty (in the case_unit) / (yield_pct / 100), plus safety_pct percent of that sum. Order the smallest whole number of cases that brings on_hand up to at least the purchase need, or 0 if on_hand already covers it. - Order total: the sum of cases × case_price over all ingredients. - Write `out/kitchen.json`: `{"dishes": {"<dish>": {"plate_cost": <dollars>, "food_cost_pct": <percent>}, ...}, "orders": {"<ingredient>": <cases>, ...}, "order_total": <dollars>}` with every ingredient in `orders`, including those needing 0 cases. ## 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.

Cost a small restaurant's synthetic menu from its recipes and turn next week's forecast into a supplier order. Recipe quantities in grams and millilitres convert to case units, trim yields raise the amount bought, plate costs are rounded once per dish, food-cost percentages are rounded to one decimal, and each ingredient is ordered in whole cases after on-hand stock and a safety margin. A chef's note gets the yield, rounding, case and safety-stock rules wrong. 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: units plate_cost food_cost ordering order_total

Best model for this task so far: Claude Haiku 5.5 (effort default (no --effort flag) · thinking on) (success, $0.0045, 16s).

Run this on another model

All runs (3)

100% pass rate across 3 eligible runs: 3 pass, 0 partial, 0 fail. Partial and failed runs are listed below with everything else.ⓘ

Runs for this use case
ModelCostTimeStep-insⓘAutonomyOutcomeTrustⓘDateCompare
AClaude Opus 5.5default · thinking on$0.1217s0Fully autonomouspassTrust 80%2026-10-24
AClaude Haiku 5.5default · thinking on$0.004516s0Fully autonomouspassTrust 80%2026-10-24
AClaude Sonnet 5.5default · thinking on$0.0613s0Fully autonomouspassTrust 80%2026-10-24

Embed this result

A live badge for this use case. It links back here and updates as new runs are published.

UseCaseRank badge for Cost a restaurant's menu dishes from recipes with trim yields and unit conversions, and build the week's supplier order in whole cases

Get new runs and ranking changes weekly

Which models got better or cheaper on real tasks like this one.