{
 "slug": "pilot",
 "kind": "pilot",
 "title": "Pilot",
 "dek": "The first test: four models, 120 moves each, before the harness was finished. Kept for the record.",
 "date": "2026-09-28",
 "x_post": null,
 "youtube": null,
 "seed": 7,
 "seeds": [
  7
 ],
 "scenario": "ladder_short",
 "method": {
  "game": "Pokémon Emerald, new game, seeded RNG, trainer ID and clock",
  "tools": "pathfinding tools (walk to a tile, talk, read, pick from menus)",
  "move_budget": 120,
  "safety_cap_usd": 1.0,
  "memory_every": 40,
  "settings": "each provider's default settings",
  "cost_basis": "cost billed by OpenRouter",
  "score": "furthest story exit from the game's own flags, then fewest moves to reach it",
  "ends": "after 120 moves, or at the safety cap, or if the model's API fails 3 moves in a row",
  "routing": "OpenRouter default provider routing",
  "harness": "These runs predate per-run version stamps. The code that plays the race hasn't changed since."
 },
 "lanes": [
  {
   "model": "anthropic/claude-sonnet-5",
   "name": "Sonnet 5",
   "maker": "Anthropic",
   "glyph": 0,
   "exit": 4,
   "exit_label": "Got a starter",
   "reached_at": 76,
   "exits": {
    "1": 4,
    "2": 15,
    "3": 76,
    "4": 76
   },
   "moves": 102,
   "cost": 1.0089,
   "outcome": "cost cap $1.00 reached",
   "no_action_moves": 0,
   "truncated_replies": 0,
   "moves_without_moving": 43,
   "tokens_in": 437963,
   "tokens_out": 11124,
   "stats": {
    "words": 1842,
    "words_per_move": 18.2,
    "think_s_avg": 4.4,
    "think_s_total": 443,
    "cost_per_move": 0.01,
    "tokens_in_per_move": 4336,
    "tokens_out_per_move": 110,
    "battle_moves": 2,
    "note_updates": 5,
    "wall_min": 11,
    "cost_curve": [
     [
      10,
      0.0842
     ],
     [
      20,
      0.1694
     ],
     [
      30,
      0.2582
     ],
     [
      40,
      0.3511
     ],
     [
      50,
      0.4586
     ],
     [
      60,
      0.5678
     ],
     [
      70,
      0.6757
     ],
     [
      80,
      0.7762
     ],
     [
      90,
      0.8709
     ],
     [
      100,
      0.9756
     ],
     [
      101,
      0.9872
     ]
    ]
   },
   "run_id": "pilot.s7.claude-sonnet-5",
   "host": "OpenRouter default routing · host not logged",
   "rank": 1
  },
  {
   "model": "deepseek/deepseek-v4-flash",
   "name": "DeepSeek V4 Flash",
   "maker": "DeepSeek",
   "glyph": 1,
   "exit": 2,
   "exit_label": "Set the clock",
   "reached_at": 92,
   "exits": {
    "1": 16,
    "2": 92
   },
   "moves": 120,
   "cost": 0.0212,
   "outcome": "step budget used up",
   "no_action_moves": 4,
   "truncated_replies": 4,
   "moves_without_moving": 25,
   "tokens_in": 280378,
   "tokens_out": 18880,
   "stats": {
    "words": 2470,
    "words_per_move": 20.6,
    "think_s_avg": 3.5,
    "think_s_total": 416,
    "cost_per_move": 0.0002,
    "tokens_in_per_move": 2336,
    "tokens_out_per_move": 157,
    "battle_moves": 0,
    "note_updates": 8,
    "wall_min": 9,
    "cost_curve": [
     [
      10,
      0.0017
     ],
     [
      20,
      0.0048
     ],
     [
      30,
      0.0068
     ],
     [
      40,
      0.0078
     ],
     [
      50,
      0.0089
     ],
     [
      60,
      0.0107
     ],
     [
      70,
      0.0119
     ],
     [
      80,
      0.0136
     ],
     [
      90,
      0.0153
     ],
     [
      100,
      0.0169
     ],
     [
      110,
      0.0187
     ],
     [
      120,
      0.0208
     ]
    ]
   },
   "run_id": "pilot.s7.deepseek-v4-flash",
   "host": "OpenRouter default routing · host not logged",
   "rank": 3
  },
  {
   "model": "google/gemini-3-flash-preview",
   "name": "Gemini 3 Flash",
   "maker": "Google",
   "glyph": 2,
   "exit": 4,
   "exit_label": "Got a starter",
   "reached_at": 99,
   "exits": {
    "1": 4,
    "2": 20,
    "3": 99,
    "4": 99
   },
   "moves": 120,
   "cost": 0.1634,
   "outcome": "step budget used up",
   "no_action_moves": 0,
   "truncated_replies": 0,
   "moves_without_moving": 45,
   "tokens_in": 278544,
   "tokens_out": 6979,
   "stats": {
    "words": 3181,
    "words_per_move": 26.5,
    "think_s_avg": 2.1,
    "think_s_total": 254,
    "cost_per_move": 0.0014,
    "tokens_in_per_move": 2321,
    "tokens_out_per_move": 58,
    "battle_moves": 2,
    "note_updates": 6,
    "wall_min": 8,
    "cost_curve": [
     [
      10,
      0.0104
     ],
     [
      20,
      0.0246
     ],
     [
      30,
      0.0375
     ],
     [
      40,
      0.0501
     ],
     [
      50,
      0.0631
     ],
     [
      60,
      0.076
     ],
     [
      70,
      0.0901
     ],
     [
      80,
      0.1037
     ],
     [
      90,
      0.1167
     ],
     [
      100,
      0.131
     ],
     [
      110,
      0.1451
     ],
     [
      120,
      0.1602
     ]
    ]
   },
   "run_id": "pilot.s7.gemini-3-flash-preview",
   "host": "OpenRouter default routing · host not logged",
   "rank": 2
  },
  {
   "model": "openai/gpt-5-mini",
   "name": "GPT-5 mini",
   "maker": "OpenAI",
   "glyph": 3,
   "exit": 1,
   "exit_label": "Left the truck",
   "reached_at": 4,
   "exits": {
    "1": 4
   },
   "moves": 120,
   "cost": 0.1674,
   "outcome": "step budget used up",
   "no_action_moves": 16,
   "truncated_replies": 0,
   "moves_without_moving": 16,
   "tokens_in": 200379,
   "tokens_out": 57419,
   "stats": {
    "words": 3345,
    "words_per_move": 27.9,
    "think_s_avg": 7.3,
    "think_s_total": 877,
    "cost_per_move": 0.0014,
    "tokens_in_per_move": 1670,
    "tokens_out_per_move": 478,
    "battle_moves": 0,
    "note_updates": 2,
    "wall_min": 17,
    "cost_curve": [
     [
      10,
      0.014
     ],
     [
      20,
      0.0252
     ],
     [
      30,
      0.0397
     ],
     [
      40,
      0.0555
     ],
     [
      50,
      0.0704
     ],
     [
      60,
      0.0839
     ],
     [
      70,
      0.097
     ],
     [
      80,
      0.1105
     ],
     [
      90,
      0.1234
     ],
     [
      100,
      0.1388
     ],
     [
      110,
      0.1511
     ],
     [
      120,
      0.1649
     ]
    ]
   },
   "run_id": "pilot.s7.gpt-5-mini",
   "host": "OpenRouter default routing · host not logged",
   "rank": 4
  }
 ],
 "moments": [
  {
   "model": "anthropic/claude-sonnet-5",
   "move": 76,
   "kind": "milestone",
   "why": "First to Exit 4: Got a starter",
   "reasoning": "The bag is open showing the Poke Ball with Torchic. I need to select it to release Torchic and save Prof. Birch."
  }
 ],
 "notes": []
}