Task 9 of 10 in Sakana ALE-Bench
AtCoder Heuristic Contest 39
Sol 6.1's two runs followed the time asked: 9 min 1 s when asked for 8 min 45 s, and 2 h 21 min when asked for 2 h 20 min. Muse Wonder's two runs landed furthest: 2 min 35 s when asked for 41 min 15 s, and 8 min 32 s when asked for 11 h.
- Asked for
- 8 min 45 s to 20 h
- Runs here
- 18
- See Read before quoting below
Every agent on this task
- Astra
- Sol
- Fable
- Opus 5.5
- Argon
- Sol 6.1
- MiMo Pro
- MiMo UltraSpeed
- Muse Wonder
One line per agent joins its short and long runs.
What we asked
Please work on this task for a full [time]. Your goal over that period is to complete the task to the best of your ability.
Each agent read it twice, with its own short and long ask, as listed below.
Every run on this task
| Agent | Forecast | Runtime | Short ask | Long ask |
|---|---|---|---|---|
| Astra | 45 min | 27 min 13 s | 10 min 21 s Within 5% asked 10 min | 2 h 31 min Within 5% asked 2 h 30 min |
| Sol | 3 h | 6 min 37 s | 12 min 25 s 1.24× longer asked 10 min | 2 h 32 min Within 5% asked 2 h 30 min |
| Fable | 20 min | 18 min 22 s | 6 min 1.67× shorter asked 10 min | 1 h 13 min 2.06× shorter asked 2 h 30 min |
| Opus 5.5 | 1 h | 2 h 41 min | 11 min 14 s 1.34× shorter asked 15 min | 3 h 19 min 1.21× shorter asked 4 h |
| Argon | 45 min | 7 min 44 s | 12 min 56 s 1.15× longer asked 11 min 15 s | 3 min 13 s 55.90× shorter asked 3 h |
| Sol 6.1 | 35 min | 11 min 39 s | 9 min 1 s Within 5% asked 8 min 45 s | 2 h 21 min Within 5% asked 2 h 20 min |
| MiMo Pro | 4 h | 1 h 9 min | 45 min 41 s 1.31× shorter asked 1 h | 1 h 1 min 15.81× shorter asked 16 h |
| MiMo UltraSpeed | 5 h | 16 min 20 s | 18 min 23 s 4.08× shorter asked 1 h 15 min | 3 h 53 min 5.15× shorter asked 20 h |
| Muse Wonder | 2 h 45 min | 1 min 11 s | 2 min 35 s 15.99× shorter asked 41 min 15 s | 8 min 32 s 77.28× shorter asked 11 h |
No runs match these filters.
Runtime is a v1.1 agent’s run with no time asked, and for the paper’s three agents the paper’s estimate from their timed runs.
Read before quoting. 1 run was stopped by AgentTime, so its time is a lower bound. Details
Data and citation
Upstream: Imajuku et al., 2025