RUN 20260725-opus5-bash-probe (design)

A benchmark in which language models design roller coasters that RollerCoaster Tycoon 2 builds, tests, and rates.

Standings

design mode — models build from scratch

modelscoreintensitynausea similaritybesttokens / cost
1st claude-opus-5 7.06 10.07 6.14 0.46 round 1/2 191 in / 150k out · $5.82

score = excitement, scaled down when similarity to a stock library design exceeds 0.5 (an exact or mirrored copy scores 0)

claude-opus-5full detail →
excitementintensity×build failed
0515intensity ceiling 10excitementintensityR1R2

Round 1 trace (61) →

excitement7.06 intensity10.07 nausea6.14
length 674 m drops 12 airtime 50 tested ok similarity 0.46 (nearest: Time Twister)
left_vertical_loopleft_corkscrew_upright_corkscrew_downright_vertical_loopleft_corkscrew_upright_corkscrew_downleft_vertical_loopleft_corkscrew_upright_corkscrew_downbrakesbrakesbrakesbrakes
81 pieces · 14 on the lift · 9 inversions
track program (81 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "end_station",
    {
      "chain": true,
      "t": "flat_to_up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_down_60",
    "down_60",
    "down_60",
    "down_60_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "flat_to_left_bank",
    "left_helix_down_large",
    "left_bank_to_flat",
    "flat_to_up_25",
    "right_vertical_loop",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25_to_flat",
    "flat_to_right_bank",
    "right_helix_up_large",
    "right_bank_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_left_bank",
    "left_helix_up_large",
    "left_bank_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "flat_to_up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat",
    "flat_to_left_bank",
    "banked_left_turn_5",
    "left_bank_to_flat",
    "flat",
    "flat",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "flat_to_left_bank",
    "banked_left_turn_5",
    "left_bank_to_flat"
  ],
  "ride_type": 51,
  "start": {
    "dir": 0,
    "x": 60,
    "y": 60
  }
}
park screenshot, claude-opus-5 round 1

Round 2 trace (64, 2 rejected) →

excitement6.96 intensity8.40 nausea4.70
length 725 m drops 13 airtime 67 tested ok similarity 0.39 (nearest: Danglefeet)
left_vertical_loopleft_corkscrew_upright_corkscrew_downleft_vertical_loopleft_corkscrew_upright_corkscrew_downleft_vertical_loopleft_corkscrew_upright_corkscrew_downleft_corkscrew_upright_corkscrew_downbrakesbrakesbrakesbrakes
96 pieces · 15 on the lift · 11 inversions · 33 added, 18 removed vs the round before (63 of 81 pieces kept)
track program (96 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "middle_station",
    "end_station",
    {
      "chain": true,
      "t": "flat_to_up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    {
      "chain": true,
      "t": "up_25"
    },
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_down_60",
    "down_60",
    "down_60",
    "down_60_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_left_bank",
    "banked_left_turn_5",
    "left_bank_to_flat",
    "flat_to_up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_left_bank",
    "banked_left_turn_5",
    "left_bank_to_flat",
    "flat_to_up_25",
    "up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_left_bank",
    "banked_left_turn_5",
    "left_bank_to_flat",
    "flat_to_up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "flat_to_up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "flat_to_right_bank",
    "banked_right_turn_5",
    "right_bank_to_flat",
    "flat_to_up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "brakes",
    "right_turn_5",
    "brakes",
    "brakes",
    "brakes",
    "right_turn_5"
  ],
  "ride_type": 51,
  "start": {
    "dir": 0,
    "x": 60,
    "y": 60
  }
}
park screenshot, claude-opus-5 round 2

← all runs