Task 4 of 18 in OSWorld
CRM leads
Astra's two runs came closest to the time asked: 26 min 14 s when asked for 15 min, and 4 h 12 min when asked for 4 h 10 min. MiMo UltraSpeed worked about the same whatever it was asked: 5 h 20 min when asked for 10 min, and 5 h 20 min when asked for 2 h 40 min.
- Asked for
- 2 min 15 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 | 48 min 8 s | 26 min 14 s 1.75× longer asked 15 min | 4 h 12 min Within 5% asked 4 h 10 min |
| Sol | 15 min | 36 min 41 s | 1 h 6 min 4.42× longer asked 15 min | 27 min 27 s 9.11× shorter asked 4 h 10 min |
| Fable | 15 min | 1 h 4 min | 14 min 41 s Within 5% asked 15 min | 1 h 15 min 3.35× shorter asked 4 h 10 min |
| Opus 5.5 | 30 min | 1 h 8 min | 6 min 28 s 1.16× shorter asked 7 min 30 s | 44 min 10 s 2.72× shorter asked 2 h |
| Argon | 9 min | 18 min 7 s | 37 min 7 s 16.50× longer asked 2 min 15 s | 38 min 49 s 1.08× longer asked 36 min |
| Sol 6.1 | 28 min | 31 min 11 s | 14 min 21 s 2.05× longer asked 7 min | 1 h 54 min Within 5% asked 1 h 52 min |
| MiMo Pro | 35 min | 6 h 9 min | 30 min 43 s 3.51× longer asked 8 min 45 s | 4 h 7 min 1.77× longer asked 2 h 20 min |
| MiMo UltraSpeed | 40 min | 10 h 4 min | 5 h 20 min 32.00× longer asked 10 min | 5 h 20 min 2.00× longer asked 2 h 40 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. 2 runs were stopped by AgentTime, so their time is a lower bound. Details
Data and citation
Upstream: Yuan et al., 2026