RUN 20260724-opus48-twister (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-4-8 6.67 7.58 4.37 0.23 round 5/6 531 in / 382k out · $26.57

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

claude-opus-4-8full detail →
excitementintensity×build failed
0515intensity ceiling 10excitementintensityR1R2R3R4R5R6

Round 1 trace (93) →

excitement6.50 intensity8.31 nausea4.48
length 502 m drops 13 airtime 97 tested ok similarity 0.35 (nearest: Fright Flight)
left_corkscrew_upright_corkscrew_downright_corkscrew_upleft_corkscrew_downleft_corkscrew_upright_corkscrew_downright_corkscrew_upleft_corkscrew_down
78 pieces · 8 inversions
track program (78 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "end_station",
    "flat_to_up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25_to_flat",
    "left_turn_5",
    "left_turn_5",
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "right_corkscrew_up",
    "left_corkscrew_down",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "right_corkscrew_up",
    "left_corkscrew_down",
    "flat_to_down_25",
    "down_25_to_flat",
    "left_turn_5",
    "left_turn_5",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat",
    "flat"
  ],
  "ride_type": 51,
  "start": {
    "dir": 2,
    "x": 50,
    "y": 45
  }
}
park screenshot, claude-opus-4-8 round 1 view 1

Round 2 trace (57, 2 rejected) →

excitement6.15 intensity6.07 nausea3.32
length 317 m drops 7 airtime 46 tested ok similarity 0.40 (nearest: Flight of the Phoenix)
left_corkscrew_upleft_corkscrew_downleft_corkscrew_upleft_corkscrew_downbrakesbrakesbrakesbrakesbrakesbrakesbrakesbrakes
60 pieces · 4 inversions · 20 added, 38 removed vs the round before (40 of 78 pieces kept)
track program (60 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "end_station",
    "flat_to_up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_60",
    "down_60",
    "down_60",
    "down_60_to_flat",
    "left_corkscrew_up",
    "left_corkscrew_down",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "left_corkscrew_up",
    "left_corkscrew_down",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes"
  ],
  "ride_type": 51,
  "start": {
    "dir": 2,
    "x": 36,
    "y": 46
  }
}
park screenshot, claude-opus-4-8 round 2 view 1

Round 3 trace (65) →

excitement5.53 intensity5.89 nausea2.98
length 259 m drops 5 airtime 35 tested ok similarity 0.38 (nearest: Hell Hound)
right_corkscrew_upright_corkscrew_down
40 pieces · 2 inversions · 7 added, 27 removed vs the round before (33 of 60 pieces kept)
track program (40 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "end_station",
    "flat_to_up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_60",
    "down_60",
    "down_60",
    "down_60_to_flat",
    "right_corkscrew_up",
    "right_corkscrew_down",
    "s_bend_left",
    "s_bend_left",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25_to_flat",
    "flat",
    "right_turn_5",
    "right_turn_5"
  ],
  "ride_type": 51,
  "start": {
    "dir": 2,
    "x": 40,
    "y": 46
  }
}
park screenshot, claude-opus-4-8 round 3 view 1

Round 4 trace (54, 1 rejected) →

excitement6.00 intensity6.22 nausea3.36
length 308 m drops 8 airtime 55 tested ok similarity 0.38 (nearest: Pendulator)
right_corkscrew_upright_corkscrew_downright_corkscrew_upright_corkscrew_downbrakesbrakesbrakesbrakes
58 pieces · 4 inversions · 24 added, 6 removed vs the round before (34 of 40 pieces kept)
track program (58 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "end_station",
    "flat_to_up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_60",
    "down_60",
    "down_60",
    "down_60_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "right_corkscrew_up",
    "right_corkscrew_down",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat_to_up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "right_corkscrew_up",
    "right_corkscrew_down",
    "brakes",
    "brakes",
    "brakes",
    "brakes"
  ],
  "ride_type": 51,
  "start": {
    "dir": 2,
    "x": 40,
    "y": 55
  }
}
park screenshot, claude-opus-4-8 round 4 view 1

Round 5 trace (86) →

excitement6.67 intensity7.58 nausea4.37
length 571 m drops 15 airtime 176 tested ok similarity 0.23 (nearest: El Condor)
left_corkscrew_upright_corkscrew_downleft_corkscrew_upright_corkscrew_downleft_corkscrew_upright_corkscrew_downleft_corkscrew_upright_corkscrew_downbrakesbrakesbrakesbrakesbrakesbrakesbrakesbrakesbrakes
90 pieces · 8 inversions · 66 added, 34 removed vs the round before (24 of 58 pieces kept)
track program (90 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "middle_station",
    "end_station",
    "flat_to_up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "right_turn_5",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat",
    "flat",
    "flat",
    "right_turn_5",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "flat",
    "flat",
    "right_turn_5",
    "brakes",
    "brakes",
    "brakes",
    "right_turn_5"
  ],
  "ride_type": 51,
  "start": {
    "dir": 2,
    "x": 30,
    "y": 40
  }
}
park screenshot, claude-opus-4-8 round 5 view 1

Round 6 trace (75, 2 rejected) →

excitement6.56 intensity6.87 nausea3.82
length 470 m drops 10 airtime 87 tested ok similarity 0.28 (nearest: El Condor)
left_corkscrew_upright_corkscrew_downleft_corkscrew_upright_corkscrew_downleft_corkscrew_upright_corkscrew_downbrakesbrakesbrakesbrakesbrakesbrakesbrakesbrakesbrakes
69 pieces · 6 inversions · 10 added, 31 removed vs the round before (59 of 90 pieces kept)
track program (69 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "middle_station",
    "end_station",
    "flat_to_up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "right_turn_5",
    "right_turn_5",
    "flat_to_down_25",
    "down_25_to_flat",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "left_corkscrew_up",
    "right_corkscrew_down",
    "s_bend_right",
    "s_bend_right",
    "s_bend_right",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "flat_to_up_60",
    "up_60_to_flat",
    "flat_to_down_60",
    "down_60_to_flat",
    "right_turn_5",
    "right_turn_5",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes",
    "brakes"
  ],
  "ride_type": 51,
  "start": {
    "dir": 0,
    "x": 60,
    "y": 45
  }
}
park screenshot, claude-opus-4-8 round 6 view 1