Task 8 of 12 in BrowseComp
BrowseComp question 8
Sol 6.1's two runs came closest to the time asked: 4 min 21 s when asked for 3 min 45 s, and 1 h 1 min when asked for 1 h. Muse Wonder worked about the same whatever it was asked: 8 min 28 s when asked for 5 min, and 4 min 35 s when asked for 1 h 20 min.
- Asked for
- 53 s 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 | 7 min | 3 min 37 s | 2 min 35 s 1.48× longer asked 1 min 45 s | 22 min 52 s 1.22× shorter asked 28 min |
| Argon | 3 min 30 s | 3 min 17 s | 1 min 19 s 1.51× longer asked 53 s | 14 min 41 s Within 5% asked 14 min |
| Sol 6.1 | 15 min | 4 min 38 s | 4 min 21 s 1.16× longer asked 3 min 45 s | 1 h 1 min Within 5% asked 1 h |
| MiMo Pro | 30 min | 11 min 16 s | 19 min 8 s 2.55× longer asked 7 min 30 s | 26 min 37 s 4.51× shorter asked 2 h |
| MiMo UltraSpeed | 42 min | 8 min 48 s | 5 min 17 s 1.99× shorter asked 10 min 30 s | 12 min 51 s 13.07× shorter asked 2 h 48 min |
| Muse Wonder | 20 min | 1 min 42 s | 8 min 28 s 1.69× longer asked 5 min | 4 min 35 s 17.46× shorter asked 1 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