Task 14 of 18 in ProgramBench
rvben/rumdl
Sol 6.1's two runs followed the time asked: 30 min 23 s when asked for 30 min, and 8 h when asked for 8 h. Muse Wonder worked about the same whatever it was asked: 12 min 18 s when asked for 4 h, and 15 min 27 s when asked for 64 h.
- Asked for
- 28 min 45 s to 100 h
- Runs here
- 12
- 6 agents
Every agent on this task
- 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 |
|---|---|---|---|---|
| Opus 5.5 | 4 h | 26 h 18 min | 54 min 56 s 1.09× shorter asked 1 h | 15 h 51 min Within 5% asked 16 h |
| Argon | 1 h 55 min | 1 h 29 min | 46 min 33 s 1.62× longer asked 28 min 45 s | 50 min 13 s 9.16× shorter asked 7 h 40 min |
| Sol 6.1 | 2 h | 58 min 8 s | 30 min 23 s Within 5% asked 30 min | 8 h Within 5% asked 8 h |
| MiMo Pro | 20 h | 1 h 46 min | 51 min 44 s 5.80× shorter asked 5 h | 3 h 6 min 25.76× shorter asked 80 h |
| MiMo UltraSpeed | 25 h | 1 h 19 min | 1 h 54 min 3.30× shorter asked 6 h 15 min | 1 h 54 min 52.43× shorter asked 100 h |
| Muse Wonder | 16 h | 18 min 52 s | 12 min 18 s 19.51× shorter asked 4 h | 15 min 27 s 248.54× shorter asked 64 h |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Yang et al., 2026