chini-014-restaurant-friday-night

Restaurant Friday Night Service

Eight tables turning every 90 minutes. Three stations on the line. The walk-in just got delivered short on prep.

Source: Restaurant operations, brigade kitchen workflow, every line cook who has been in the weeds

Prompt

Design the front-of-house plus kitchen workflow for a 40-seat dinner restaurant on Friday night.

Functional:
- Hosts seat parties at 12 tables. Server takes order, fires it to the kitchen.
- Kitchen splits orders by station: cold (salads, starters), hot (entrees), pastry (dessert). Each station has one cook.
- Expediter assembles plates from stations, calls server for runner pickup.
- Bar is parallel: cocktails fired direct from server, no kitchen path.

Non-functional:
- A 7pm rush (4x normal arrival) must not blow ticket times past 25 minutes for entrees.
- If hot station falls behind, expediter must pace the cold/pastry stations to stay synchronized. Plates cannot be held under heat lamp more than 5 minutes.
- If walk-in shorted on prep (one entree 86'd), server must be notified to update guests before food fires, not at pickup.

Return a Chinilla CanvasState. Components: hosts, servers, line cooks, expediter, bar. Behaviors: queue (ticket rail), split (station routing), batch (table-by-table coursing), circuitbreaker (86'd item handling), ratelimit (table turn pace).

Constraints

Max components: 14
Required behaviors: queue, split, circuitbreaker
Monthly budget: $75000

Stress scenarios

Steady service

baseline

Tables turning at normal pace. All stations operating.

7pm rush

spike

Arrivals 4x baseline. Kitchen must absorb without lamp-time violations.

Halibut 86'd mid-service

outage

Hot station out of a key entree. Server must intercept tickets before they fire.

Hot station behind

latency

Hot station ticket time spikes. Expediter must pace cold/pastry to stay synced.

Pass criteria (overall)

Min stability score: 60
Max drop rate: 10.0%
Min delivery rate: 85.0%
Max errors: 8

Submit your run

Submissions go through the chini-bench CLI. It calls your model with your key, scores the result locally, and posts to the leaderboard. Nothing leaves your machine except the canvas it produces.

End-to-end:

pip install git+https://github.com/collapseindex/chini-bench-cli.git
export OPENROUTER_API_KEY=...

chini-bench run chini-014-restaurant-friday-night \
  --provider openrouter --model google/gemini-2.0-flash-001 \
  --as alice --x alice --linkedin alice-builds

Or inspect the prompt first:

chini-bench prompt chini-014-restaurant-friday-night

Providers: openai · anthropic · google · openrouter · ollama

Leaderboard

Rank	Submitter	Model	Score	Stability	Delivery	Design	Pass	Links
#1	alex default	X x-ai/grok-4.20	92	83.0	100.0	100.0	✓	X
#2	alex default	O openai/gpt-5.4	27	0.0	19.0	100.0	✗	X
#3	alex default	A anthropic/claude-sonnet-4.6	20	0.0	0.0	75.0	✗	X
#4	alex default	G google/gemini-3.1-pro-preview	20	0.0	0.0	75.0	✗	X

Per-scenario breakdown of the top run

Scenario	Health	Drop rate	Delivered	Pass
baseline	83.0	1.8%	493	✓
rush	86.0	1.5%	1836	✓
item-86	77.0	1.1%	364	✓
slow-cook	84.0	1.0%	472	✓

How is this scored? →