HUSTLE SIM BENCHMARK · LOCAL, THINKING LOW

gpt-oss 20B

Openai · open weights · released 2025-08-05 · route gpt-oss:20b · Model profile · Chess results · Sudoku results · Mines results · FreeCell results · Nations Royale results · Elemental Armory results · Micro Mouse results

Results

Score
1295%: 0 to 44 · 3 of 6 base runs so far
Reached the goal in time
3 of 3 runs
Ran out of replies / forfeited / never
0 / 0 / 0
Game time over the best line's
11.0x
As fast as the best line or faster
0 of 3 runs
Decisions per run
21.3
Actions per decision
6.05
Purchases the game refused
96 of 387 (24.8%)
Decisions with no readable actions
0.0%
Retries for a reply with no actions tag
0
Tokens per decision (in / out)
2,357 / 3,658
Reasoning tokens per decision
n/a
Model minutes per run
135.9
Runs on other rules (not counted)
0
API-equivalent cost per run
n/a
Seconds per decision
382.2
Replies cut off at the length limit
0.0%
Aborted on our side (not counted)
0

By tier

TierGoalRepliesRunsScoreMedian time vs bestIn time / ran out / never
Fresh start$1M30155h 37m vs 17m 30s1 / 0 / 0
Takeover$1B601271d 7h vs 8h 24m1 / 0 / 0
Empire$1T8015108d 4h vs 5d 20h1 / 0 / 0

Against the game's strategies, run by run

Runs where it reached the goal sooner than each strategy did from the same start.

StrategyFaster
Best income per dollar0 of 3
Deepen first0 of 3
Cheapest thing available0 of 3
Chase speed milestones0 of 3
Perks before levels2 of 3
Automate everything3 of 3
Collect first3 of 3
Unguided (seed 3)3 of 3
Unguided (seed 1)3 of 3
Unguided (seed 2)3 of 3

Every run