MAAS_QWEN-3-8-27B · 20260915-qwen38-27b-library-long

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

excitement 5.85 intensity 5.16 nausea 2.45 length 410 m drops 2 airtime 4 pieces 70
maas_qwen-3-8-27b · 1st of 1 compare all models →

library · steel twister · opencode · library mode — models may search the stock track design library, which measures retrieval and adaptation; copies score zero

excitementintensity×build failed
0515intensity ceiling 10excitementintensity×round 1: build failed (0/0 pieces placed): finish_and_test: no active ride; call new_ride firstR1×round 2: build failed (50/50 pieces placed): finish_and_test: set_status(testing): Track is not a complete circuit [track starts at {"bank":0,"dir":2,"slope":0,"x":45,"y":55,"z":112} and currently ends at {"bank":0,"dir":2,"slope":0,"x":45,"y":55,"z":128}; a closed circuit requires these to match]R2R3R4R5×round 6: build failed (58/58 pieces placed): finish_and_test: set_status(testing): Track is not a complete circuit [track starts at {"bank":0,"dir":1,"slope":0,"x":40,"y":40,"z":112} and currently ends at {"bank":0,"dir":3,"slope":0,"x":50,"y":34,"z":112}; a closed circuit requires these to match]R6
score
5.85
intensity
5.16
nausea
2.45
best round
5
rated rounds
3/6
similarity
0.43
tokens
6.9M fresh in · 0 cache reads · 931k out
cost
$0.00
6 round(s) · per-round detail
chain lift inversion brake / booster round before height is the track's real elevation, walked from the piece list

Round 1 build failed trace (83, 3 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, maas_qwen-3-8-27b round 1

Round 2 build failed trace (77) →

build failed (50/50 pieces placed): finish_and_test: set_status(testing): Track is not a complete circuit [track starts at {"bank":0,"dir":2,"slope":0,"x":45,"y":55,"z":112} and currently ends at {"bank":0,"dir":2,"slope":0,"x":45,"y":55,"z":128}; a closed circuit requires these to match]

50 pieces · 8 on the lift
track program (50 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"
    },
    "up_25_to_flat",
    "flat",
    "flat",
    "left_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "left_turn_5",
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "left_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "left_turn_5"
  ],
  "ride_type": 51,
  "start": {
    "dir": 2,
    "x": 45,
    "y": 55
  }
}

download scored .park

replay

Round 3 verified circuit (44 pieces) trace (197, 7 rejected) →

excitement0.63 intensity0.32 nausea0.15
length 244 m drops 1 airtime 0 tested ok similarity 0.50 (nearest: Rocket Rush)
44 pieces · 3 on the lift · 12 added, 18 removed vs the round before (32 of 50 pieces kept)
track program (44 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"
    },
    "up_25_to_flat",
    "flat_to_down_25",
    "down_25",
    "down_25",
    "down_25",
    "down_25_to_flat",
    "right_turn_5",
    "flat",
    "flat",
    "flat",
    "right_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "right_turn_5",
    "flat",
    "flat",
    "flat",
    "right_turn_5",
    "flat",
    "flat"
  ],
  "ride_type": 51,
  "start": {
    "dir": 0,
    "x": 58,
    "y": 60
  }
}

download scored .park

replay

Round 4 verified circuit (72 pieces) trace (158, 2 rejected) →

excitement1.36 intensity0.87 nausea0.39
length 369 m drops 1 airtime 0 tested ok similarity 0.46 (nearest: King Rapids)
72 pieces · 8 on the lift · 33 added, 5 removed vs the round before (39 of 44 pieces kept)
track program (72 pieces)
{
  "pieces": [
    "begin_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"
    },
    "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_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "left_turn_5",
    "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",
    "left_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "left_turn_5",
    "flat",
    "flat"
  ],
  "ride_type": 51,
  "start": {
    "dir": 1,
    "x": 40,
    "y": 45
  }
}

download scored .park

replay

Round 5 verified circuit (70 pieces) trace (209, 2 rejected) →

excitement5.85 intensity5.16 nausea2.45
length 410 m drops 2 airtime 4 tested ok similarity 0.43 (nearest: Hover Bother)
right_vertical_loop
70 pieces · 12 on the lift · 1 inversion · 14 added, 16 removed vs the round before (56 of 72 pieces kept)
track program (70 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",
    "down_60_to_down_25",
    "down_25_to_flat",
    "flat_to_down_25",
    "down_25_to_flat",
    "left_turn_5",
    "flat",
    "flat_to_up_25",
    "right_vertical_loop",
    "down_25_to_flat",
    "left_turn_5",
    "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",
    "flat",
    "left_turn_5",
    "flat",
    "flat",
    "flat",
    "flat",
    "flat",
    "left_turn_5",
    "flat",
    "flat"
  ],
  "ride_type": 51,
  "start": {
    "dir": 1,
    "x": 34,
    "y": 44
  }
}

download scored .park

replay

Round 6 build failed trace (122, 1 rejected) →

build failed (58/58 pieces placed): finish_and_test: set_status(testing): Track is not a complete circuit [track starts at {"bank":0,"dir":1,"slope":0,"x":40,"y":40,"z":112} and currently ends at {"bank":0,"dir":3,"slope":0,"x":50,"y":34,"z":112}; a closed circuit requires these to match]

58 pieces · 11 on the lift · 4 added, 16 removed vs the round before (54 of 70 pieces kept)
track program (58 pieces)
{
  "pieces": [
    "begin_station",
    "middle_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"
    },
    "up_25_to_flat",
    "flat_to_down_60",
    "down_60",
    "down_60",
    "down_60_to_down_25",
    "down_25_to_flat",
    "flat",
    "flat",
    "flat",
    "right_turn_5",
    "right_turn_5",
    "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",
    "left_turn_5",
    "right_turn_5"
  ],
  "ride_type": 51,
  "start": {
    "dir": 1,
    "x": 40,
    "y": 40
  }
}

download scored .park

replay

← 20260915-qwen38-27b-library-long · all runs