Task 9 of 18 in OSWorld
MuseScore transcription
Astra's two runs came closest to the time asked: 19 min 6 s when asked for 15 min, and 4 h 12 min when asked for 4 h 10 min. Sol worked about the same whatever it was asked: 1 h 5 min when asked for 15 min, and 45 min 30 s when asked for 4 h 10 min.
- Asked for
- 5 min to 10 h
- Runs here
- 16
- 8 agents
Every agent on this task
- Astra
- Sol
- Fable
- Opus 5.5
- Argon
- Sol 6.1
- MiMo Pro
- MiMo UltraSpeed
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 | 2 h | 1 h 14 min | 19 min 6 s 1.27× longer asked 15 min | 4 h 12 min Within 5% asked 4 h 10 min |
| Sol | 1 h | 46 min 11 s | 1 h 5 min 4.33× longer asked 15 min | 45 min 30 s 5.49× shorter asked 4 h 10 min |
| Fable | 1 h | 38 min 40 s | 11 min 46 s 1.27× shorter asked 15 min | 2 h 38 min 1.58× shorter asked 4 h 10 min |
| Opus 5.5 | 1 h 20 min | 4 h 25 min | 31 min 22 s 1.57× longer asked 20 min | 6 h 3 min 1.13× longer asked 5 h 20 min |
| Argon | 20 min | 21 min 42 s | 39 min 1 s 7.80× longer asked 5 min | 43 min 11 s 1.85× shorter asked 1 h 20 min |
| Sol 6.1 | 1 h 30 min | 1 h 5 min | 22 min 48 s Within 5% asked 22 min 30 s | 4 h 14 min 1.42× shorter asked 6 h |
| MiMo Pro | 2 h 30 min | 4 h 4 min | 44 min 22 s 1.18× longer asked 37 min 30 s | 6 h 6 min 1.64× shorter asked 10 h |
| MiMo UltraSpeed | 1 h 30 min | 2 h 9 min | 21 min 59 s Within 5% asked 22 min 30 s | 4 h 16 min 1.41× shorter asked 6 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: Yuan et al., 2026