RUN 20260731-deepseek-v4-flash-design (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 openrouter_deepseek_deepseek-v4-flash-0731 7.29 9.33 4.89 0.32 round 6/6 2.4M in / 291k out · $0.41

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

openrouter_deepseek_deepseek-v4-flash-0731full detail →
excitementintensity×build failed
0515intensity ceiling 10excitementintensityR1R2×round 3: build failed (0/0 pieces placed): finish_and_test: no active ride; call new_ride firstR3×round 4: build failed (21/21 pieces placed): finish_and_test: set_status(testing): Track is not a complete circuit [track starts at {"bank":0,"dir":2,"slope":0,"x":30,"y":90,"z":112} and currently ends at {"bank":15,"dir":3,"slope":0,"x":51,"y":88,"z":208}; a closed circuit requires these to match]R4R5R6

Round 1 verified circuit (86 pieces) trace (45) →

excitement1.38 intensity0.96 nausea0.44
length 398 m drops 1 airtime 0 tested ok similarity 0.51 (nearest: Hover Bother) penalized score 1.35
86 pieces · 8 on the lift
track program (86 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_to_flat"
    },
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "right_turn_3",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "right_turn_3",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "right_turn_3",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "right_turn_3",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat"
  ],
  "ride_type": 51,
  "start": {
    "dir": 0,
    "x": 40,
    "y": 80
  }
}

download scored .park

park screenshot, openrouter_deepseek_deepseek-v4-flash-0731 round 1

Round 2 verified circuit (42 pieces) trace (33, 1 rejected) →

excitement2.98 intensity11.91 nausea7.68
length 225 m drops 3 airtime 18 tested ok similarity 0.45 (nearest: Doubledrop)
right_corkscrew_upright_corkscrew_downboosterboosterbrakes
42 pieces · 8 on the lift · 2 inversions · 17 added, 61 removed vs the round before (25 of 86 pieces kept)
track program (42 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_to_flat"
    },
    "flat_to_down_60",
    "down_60",
    "down_60_to_flat",
    "flat",
    "flat",
    "right_corkscrew_up",
    "right_corkscrew_down",
    "flat",
    "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",
    {
      "speed": 20,
      "t": "booster"
    },
    "flat",
    "flat",
    {
      "speed": 30,
      "t": "booster"
    },
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "right_turn_3",
    "right_turn_3",
    "flat",
    {
      "speed": 8,
      "t": "brakes"
    }
  ],
  "ride_type": 51,
  "start": {
    "dir": 0,
    "x": 50,
    "y": 85
  }
}

download scored .park

park screenshot, openrouter_deepseek_deepseek-v4-flash-0731 round 2

Round 3 build failed trace (5, 1 rejected) →

build failed (0/0 pieces placed): finish_and_test: no active ride; call new_ride first

track program (0 pieces)
{
  "pieces": [],
  "ride_type": 51,
  "start": {
    "dir": 0,
    "x": 0,
    "y": 0
  }
}

download scored .park

park screenshot, openrouter_deepseek_deepseek-v4-flash-0731 round 3

Round 4 build failed trace (29, 1 rejected) →

build failed (21/21 pieces placed): finish_and_test: set_status(testing): Track is not a complete circuit [track starts at {"bank":0,"dir":2,"slope":0,"x":30,"y":90,"z":112} and currently ends at {"bank":15,"dir":3,"slope":0,"x":51,"y":88,"z":208}; a closed circuit requires these to match]

right_corkscrew_up
21 pieces · 7 on the lift · 1 inversion
track program (21 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "end_station",
    "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"
    },
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "right_corkscrew_up"
  ],
  "ride_type": 51,
  "start": {
    "dir": 2,
    "x": 30,
    "y": 90
  }
}

download scored .park

park screenshot, openrouter_deepseek_deepseek-v4-flash-0731 round 4

Round 5 verified circuit (92 pieces) trace (179, 2 rejected) →

excitement6.41 intensity6.22 nausea3.07
length 576 m drops 4 airtime 0 tested ok similarity 0.42 (nearest: Hover Bother)
boosterleft_vertical_loopleft_vertical_loopboosterboosterright_vertical_loopbooster
92 pieces · 7 on the lift · 3 inversions · 79 added, 8 removed vs the round before (13 of 21 pieces kept)
track program (92 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "middle_station",
    "middle_station",
    "end_station",
    "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"
    },
    "up_25_to_flat",
    "flat_to_down_60",
    "down_60",
    "down_60_to_down_25",
    "down_25_to_flat",
    "right_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat_to_right_bank",
    "banked_right_turn_5",
    "right_bank_to_flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    {
      "speed": 25,
      "t": "booster"
    },
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    {
      "speed": 25,
      "t": "booster"
    },
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "right_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    {
      "speed": 25,
      "t": "booster"
    },
    "flat_to_up_25",
    "right_vertical_loop",
    "down_25_to_flat",
    {
      "speed": 25,
      "t": "booster"
    },
    "flat",
    "right_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat"
  ],
  "ride_type": 51,
  "start": {
    "dir": 0,
    "x": 50,
    "y": 60
  }
}

download scored .park

park screenshot, openrouter_deepseek_deepseek-v4-flash-0731 round 5

Round 6 verified circuit (91 pieces) trace (120, 3 rejected) →

Azure Fang track rails supports
excitement7.29 intensity9.33 nausea4.89
length 783 m drops 11 airtime 22 tested ok similarity 0.32 (nearest: Rotovator)
boosterleft_vertical_loopright_vertical_loopboosterleft_vertical_loopright_vertical_loopboosterleft_vertical_loopright_vertical_loopboosterboosterleft_vertical_loopright_vertical_loopboosterbrakesbrakes
91 pieces · 12 on the lift · 8 inversions · 49 added, 50 removed vs the round before (42 of 92 pieces kept)
track program (91 pieces)
{
  "pieces": [
    "begin_station",
    "middle_station",
    "middle_station",
    "middle_station",
    "end_station",
    "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"
    },
    "up_25_to_flat",
    "flat_to_down_60",
    "down_60",
    "down_60_to_down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "flat",
    {
      "speed": 25,
      "t": "booster"
    },
    "flat",
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_up_25",
    "right_vertical_loop",
    "down_25_to_flat",
    "flat",
    {
      "speed": 25,
      "t": "booster"
    },
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_up_25",
    "right_vertical_loop",
    "down_25_to_flat",
    "left_turn_5",
    "left_turn_5",
    "s_bend_left",
    "flat",
    "flat",
    {
      "speed": 25,
      "t": "booster"
    },
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_up_25",
    "right_vertical_loop",
    "down_25_to_flat",
    "flat",
    {
      "speed": 25,
      "t": "booster"
    },
    "flat",
    "flat_to_up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25_to_flat",
    "flat",
    {
      "speed": 20,
      "t": "booster"
    },
    "flat_to_up_25",
    "left_vertical_loop",
    "down_25_to_flat",
    "flat_to_up_25",
    "right_vertical_loop",
    "down_25_to_flat",
    "flat",
    {
      "speed": 20,
      "t": "booster"
    },
    "flat",
    "flat_to_up_25",
    "up_25",
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25_to_flat",
    "flat",
    "flat",
    "flat",
    {
      "speed": 8,
      "t": "brakes"
    },
    {
      "speed": 8,
      "t": "brakes"
    },
    "flat",
    "left_turn_5",
    "left_turn_3"
  ],
  "ride_type": 51,
  "start": {
    "dir": 2,
    "x": 40,
    "y": 80
  }
}

download scored .park

park screenshot, openrouter_deepseek_deepseek-v4-flash-0731 round 6

← all runs