Task 1 of 18 in OSWorld
Defense schedule
Sol 6.1's two runs followed the time asked: 6 min 23 s when asked for 6 min 15 s, and 1 h 41 min when asked for 1 h 40 min. MiMo Pro worked about the same whatever it was asked: 1 h 47 min when asked for 3 min 30 s, and 1 h 52 min when asked for 56 min.
- Asked for
- 2 min 30 s to 4 h 10 min
- Runs here
- 16
- See Read before quoting below
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 | 15 min | 38 min 12 s | 16 min 34 s 1.10× longer asked 15 min | 4 h 12 min Within 5% asked 4 h 10 min |
| Sol | 10 min | 29 min 53 s | 1 h 2 min 4.14× longer asked 15 min | 20 min 2 s 12.48× shorter asked 4 h 10 min |
| Fable | 20 min | 30 min 53 s | 30 min 13 s 2.01× longer asked 15 min | 54 min 46 s 4.57× shorter asked 4 h 10 min |
| Opus 5.5 | 27 min | 33 min 30 s | 17 min 41 s 2.62× longer asked 6 min 45 s | 43 min 35 s 2.48× shorter asked 1 h 48 min |
| Argon | 10 min | 23 min 40 s | 15 min 53 s 6.36× longer asked 2 min 30 s | 45 min 36 s 1.14× longer asked 40 min |
| Sol 6.1 | 25 min | 9 min 51 s | 6 min 23 s Within 5% asked 6 min 15 s | 1 h 41 min Within 5% asked 1 h 40 min |
| MiMo Pro | 14 min | 2 h 34 min | 1 h 47 min 30.58× longer asked 3 min 30 s | 1 h 52 min 2.00× longer asked 56 min |
| MiMo UltraSpeed | 35 min | 2 h 31 min | 3 h 20 min 22.90× longer asked 8 min 45 s | 3 h 1.28× longer asked 2 h 20 min |
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: Yuan et al., 2026