Task 3 of 12 in BrowseComp
BrowseComp question 3
Sol 6.1's two runs followed the time asked: 6 min 10 s when asked for 6 min, and 1 h 37 min when asked for 1 h 36 min. MiMo Pro worked about the same whatever it was asked: 33 min 6 s when asked for 17 min 30 s, and 17 min 31 s when asked for 4 h 40 min.
- Asked for
- 1 min 30 s to 4 h 40 min
- 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 | 20 min | 1 h 47 min | 58 min 10 s 11.63× longer asked 5 min | 1 h 9 min 1.15× shorter asked 1 h 20 min |
| Argon | 6 min | 22 min 12 s | 1 min 43 s 1.14× longer asked 1 min 30 s | 24 min 18 s Within 5% asked 24 min |
| Sol 6.1 | 24 min | 14 min 30 s | 6 min 10 s Within 5% asked 6 min | 1 h 37 min Within 5% asked 1 h 36 min |
| MiMo Pro | 1 h 10 min | 6 min 28 s | 33 min 6 s 1.89× longer asked 17 min 30 s | 17 min 31 s 15.98× shorter asked 4 h 40 min |
| MiMo UltraSpeed | 38 min | 15 min 19 s | 12 min 7 s 1.28× longer asked 9 min 30 s | 13 min 14 s 11.48× shorter asked 2 h 32 min |
| Muse Wonder | 24 min | 20 min 26 s | 16 min 13 s 2.70× longer asked 6 min | 21 min 36 s 4.45× shorter asked 1 h 36 min |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Wei et al., 2025