Task 11 of 18 in ProgramBench
paradigmxyz/solar
Sol 6.1's two runs followed the time asked: 22 min 59 s when asked for 22 min 30 s, and 6 h when asked for 6 h. Muse Wonder worked about the same whatever it was asked: 22 min 51 s when asked for 3 h 30 min, and 15 min 38 s when asked for 56 h.
- Asked for
- 18 min 45 s to 80 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 | 3 h 30 min | 44 h 34 min | 14 min 52 s 3.53× shorter asked 52 min 30 s | 13 h 22 min Within 5% asked 14 h |
| Argon | 1 h 15 min | 2 h 51 min | 18 min Within 5% asked 18 min 45 s | 4 h 34 min 1.10× shorter asked 5 h |
| Sol 6.1 | 1 h 30 min | 29 min 42 s | 22 min 59 s Within 5% asked 22 min 30 s | 6 h Within 5% asked 6 h |
| MiMo Pro | 20 h | 1 h 25 min | 1 h 30 min 3.35× shorter asked 5 h | 1 h 11 min 67.21× shorter asked 80 h |
| MiMo UltraSpeed | 15 h | 33 min 38 s | 57 min 16 s 3.93× shorter asked 3 h 45 min | 2 h 26 min 24.65× shorter asked 60 h |
| Muse Wonder | 14 h | 15 min 4 s | 22 min 51 s 9.19× shorter asked 3 h 30 min | 15 min 38 s 214.83× shorter asked 56 h |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Yang et al., 2026