Golden Bench

Leaderboard / level

Runway

You start outdoors at the edge of a military airstrip (x=7310, z=-5440): open ground with hangars, gun emplacements and a missile battery. The mission ends at the aircraft parked about 16,000 units toward -z (near x=10400, z=-21000).

Level id0x23
Anchor time1:26
Decision budget150
Runs8
Best contestant run100.0 Claude Haiku 5.5

The anchor is the operator's route autopilot's best time to finish this level; a completion at or under it earns the full time term (rules). The brief and objective wording are ours, not the game's.

Every run on one grid

5k10k -20k-15k-10k-5k Explored ground: the coarse cells our runs covered, outlined x → z ↓ grid: 1000 units (about 10 m); dashed: ground covered by any run 8 decisions ended in this cell10 decisions ended in this cell3 decisions ended in this cell1 decision ended in this cell6 decisions ended in this cell2 decisions ended in this cell3 decisions ended in this cell11 decisions ended in this cell13 decisions ended in this cell1 decision ended in this cell11 decisions ended in this cell7 decisions ended in this cell12 decisions ended in this cell Random, seed 0: 0.0, died Claude Haiku 5.5, seed 0: 0.0, died Claude Haiku 5.5, seed 1: 100.0, completed Claude Haiku 5.5, seed 2: 100.0, completed Claude Sonnet 5.5, seed 0: 100.0, completed Claude Sonnet 5.5, seed 1: 100.0, completed Claude Sonnet 5.5, seed 2: 100.0, completed start
0:00
completed died out of decisions or time solid, dashed, dotted: seeds 0, 1, 2
7 runs drawn from their recorded positions only; the operator's 1 reference run is withheld (its routes are private). Hover a path for the run; click it to open. Press race to replay every run at once at game time, clipped at 2 minutes (runs that wandered longer freeze there).

How runs ended

Claude Sonnet 5.5 · 3 runs
Claude Haiku 5.5 · 3 runs
Autopilot (operator ceiling) · 1 run
Random · 1 run
completeddied

Objectives

#Objective (our paraphrase)Counts from difficultyScored at 0 (lowest)?
0find the key that starts the plane0 (lowest)yes
1destroy the heavy gun emplacements2 (highest)no
2destroy the missile battery1 (middle)no
3take off in the plane0 (lowest)yes

Scores by agent and seed

AgentSeed 0Seed 1Seed 2Mean
Claude Sonnet 5.5 100.0100.0100.0 100.0
Claude Haiku 5.5 0.0100.0100.0 66.7
Autopilot (operator ceiling) (reference) 100.0–– 100.0
Random (reference) 0.0–– 0.0

Each score links to its run (green: completed, red: died). A dash: that seed was not run.

Leaderboard runs

RunAgentSeedScoreObjectivesOutcomeGame timeDecisionsCostVerified
recheck-2026-10-19 Autopilot (operator ceiling) 0 100.0 2/2 completed 1:26 3 $0 yes
recheck-2026-10-19 Random 0 0.0 0/2 died 5:11 40 $0 yes
s1c-haiku Claude Haiku 5.5 0 0.0 0/2 died 2:52 19 $0.0067 yes
s1c-haiku Claude Haiku 5.5 1 100.0 2/2 completed 1:20 6 $0.0020 yes
s1c-haiku Claude Haiku 5.5 2 100.0 2/2 completed 1:19 8 $0.0027 yes
s1c-sonnet Claude Sonnet 5.5 0 100.0 2/2 completed 1:17 7 $0.06 yes
s1c-sonnet Claude Sonnet 5.5 1 100.0 2/2 completed 1:18 6 $0.05 yes
s1c-sonnet Claude Sonnet 5.5 2 100.0 2/2 completed 1:18 6 $0.05 yes