Task 4 of 12 in BrowseComp
BrowseComp question 4
Sol 6.1's two runs came closest to the time asked: 5 min 19 s when asked for 5 min, and 1 h 23 min when asked for 1 h 20 min. MiMo UltraSpeed worked about the same whatever it was asked: 5 min 34 s when asked for 10 min 30 s, and 6 min 6 s when asked for 2 h 48 min.
- Asked for
- 2 min to 2 h 48 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 | 15 min | 56 min 9 s | 13 min 11 s 3.52× longer asked 3 min 45 s | 1 h Within 5% asked 1 h |
| Argon | 8 min | 36 min 22 s | 7 min 52 s 3.93× longer asked 2 min | 32 min 21 s Within 5% asked 32 min |
| Sol 6.1 | 20 min | 4 min 31 s | 5 min 19 s 1.06× longer asked 5 min | 1 h 23 min Within 5% asked 1 h 20 min |
| MiMo Pro | 35 min | 14 min 13 s | 33 min 29 s 3.83× longer asked 8 min 45 s | 29 min 9 s 4.80× shorter asked 2 h 20 min |
| MiMo UltraSpeed | 42 min | 15 min 56 s | 5 min 34 s 1.89× shorter asked 10 min 30 s | 6 min 6 s 27.58× shorter asked 2 h 48 min |
| Muse Wonder | 35 min | 30 min 14 s | 10 min 26 s 1.19× longer asked 8 min 45 s | 12 min 34 s 11.14× shorter asked 2 h 20 min |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Wei et al., 2025