Enclosure build up
hard tier · iterate
The requests, in order
1Make a solid rectangular block 80 mm by 60 mm by 30 mm, lying in the XY plane with Z up: the 80 mm side along X, the 60 mm side along Y, centred on the origin in X and Y, and its bottom face on Z = 0. Name the sub-part exactly `box`.
2Hollow it into an open-top enclosure with 2 mm walls and a 2 mm floor, keeping the outer size exactly as it is. Then add four standoffs standing on the inside of the floor: each 6 mm in diameter and 6 mm tall, at X = ±30 and Y = ±20, each with a 2.5 mm hole 5 mm deep down from its own top face.
3Cut a cable notch into the +X wall: 12 mm wide centred on Y = 0, running from the top edge down 8 mm. Keep everything else exactly as it is.
What the AI judge checks
- An open-topped rectangular box with thin walls and a floor
- Four short round standoffs standing up from the inside of the floor
- Each standoff has a hole down its centre
- A rectangular notch cut down into the top edge of one of the shorter end walls
- Nothing inside or outside the box beyond the standoffs and the notch
Measured checks
- one sub-part named box (gate)
- one sub-part named box (gate)
- outer size still 80 × 60 × 30 (gate)
- standoff at (30, 20) still stands 6 mm tall on the floor (gate)
- standoff at (30, -20) still stands 6 mm tall on the floor (gate)
- standoff at (-30, 20) still stands 6 mm tall on the floor (gate)
- standoff at (-30, -20) still stands 6 mm tall on the floor (gate)
- hollowed to a 2 mm shell, not still solid
- one sub-part named box (gate)
- outer size still 80 × 60 × 30 (gate)
- standoff at (30, 20) still stands 6 mm tall on the floor (gate)
- standoff at (30, -20) still stands 6 mm tall on the floor (gate)
- standoff at (-30, 20) still stands 6 mm tall on the floor (gate)
- standoff at (-30, -20) still stands 6 mm tall on the floor (gate)
- notch is clear at the top of the +X wall
- wall beside the notch is intact
- still a hollow shell after the notch
Ranked by hand
One reviewer put one attempt from each model in order, best first, ties left tied.
- #1 tieOpus 5.5

- #1 tieGemini 3.8 Flash

- #1 tieDeepSeek V4.1 Flash

- #1 tieGPT-6 Sol

- #5GPT-6 Luna

- #6Kimi K3 (Moonshot)
Ordering by hand decides Human rank and Human rate on the leaderboard; it never affects Overall or Strict.
Every model's attempts
This is an older run. See the current results.
Run hard-baseline-2026-09-232026-09-23commit 6edf19e0spend $212.67 + $43.09 AI judging, 15 unpricedstopped early (interrupted)Other runs: 2026-10-16T18-49-10-hard-compact-2026-10-16, 2026-09-10T19-44-23-baseline-2026-09, 2026-09-04T20-02-10-stage2-hard-cases, 2026-09-04T19-48-31-screen-14-fixed