Task 12 of 12 in BrowseComp
BrowseComp question 12
Sol 6.1's two runs came closest to the time asked: 3 min 17 s when asked for 3 min, and 48 min 53 s when asked for 48 min. MiMo UltraSpeed worked about the same whatever it was asked: 5 min 33 s when asked for 11 min 15 s, and 9 min 38 s when asked for 3 h.
- Asked for
- 1 min to 3 h
- 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 | 12 min | 9 min 5 s | 9 min 49 s 3.27× longer asked 3 min | 56 min 57 s 1.19× longer asked 48 min |
| Argon | 4 min | 4 min 57 s | 1 min 28 s 1.47× longer asked 1 min | 19 min 6 s 1.19× longer asked 16 min |
| Sol 6.1 | 12 min | 3 min 34 s | 3 min 17 s 1.10× longer asked 3 min | 48 min 53 s Within 5% asked 48 min |
| MiMo Pro | 45 min | 32 min 2 s | 45 min 37 s 4.05× longer asked 11 min 15 s | 58 min 51 s 3.06× shorter asked 3 h |
| MiMo UltraSpeed | 45 min | 8 min 13 s | 5 min 33 s 2.03× shorter asked 11 min 15 s | 9 min 38 s 18.69× shorter asked 3 h |
| Muse Wonder | 35 min | 8 min 19 s | 5 min 57 s 1.47× shorter asked 8 min 45 s | 18 min 3 s 7.75× 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