Task 1 of 10 in Sakana ALE-Bench
AtCoder Heuristic Contest 7
Astra's two runs followed the time asked: 10 min 10 s when asked for 10 min, and 2 h 31 min when asked for 2 h 30 min. Muse Wonder's two runs landed furthest: 8 min 42 s when asked for 30 min, and 4 min 6 s when asked for 8 h.
- Asked for
- 6 min 15 s to 15 h
- Runs here
- 18
- 9 agents
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 | 30 min | 26 min 53 s | 10 min 10 s Within 5% asked 10 min | 2 h 31 min Within 5% asked 2 h 30 min |
| Sol | 1 h | 26 min 26 s | 12 min 52 s 1.29× longer asked 10 min | 2 h 31 min Within 5% asked 2 h 30 min |
| Fable | 15 min | 26 min 39 s | 10 min 4 s Within 5% asked 10 min | 1 h 4 min 2.36× shorter asked 2 h 30 min |
| Opus 5.5 | 45 min | 2 h 43 min | 10 min 1 s 1.12× shorter asked 11 min 15 s | 2 h 25 min 1.24× shorter asked 3 h |
| Argon | 25 min | 1 h 40 min | 10 min 28 s 1.67× longer asked 6 min 15 s | 1 h 36 min Within 5% asked 1 h 40 min |
| Sol 6.1 | 35 min | 12 min 17 s | 9 min 3 s Within 5% asked 8 min 45 s | 2 h 22 min Within 5% asked 2 h 20 min |
| MiMo Pro | 3 h 45 min | 1 h 53 min | 1 h 6 min 1.17× longer asked 56 min 15 s | 2 h 54 min 5.18× shorter asked 15 h |
| MiMo UltraSpeed | 2 h | 29 min 41 s | 31 min 9 s Within 5% asked 30 min | 1 h 56 min 4.13× shorter asked 8 h |
| Muse Wonder | 2 h | 28 min 40 s | 8 min 42 s 3.45× shorter asked 30 min | 4 min 6 s 117.07× shorter asked 8 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.
Data and citation
Upstream: Imajuku et al., 2025