Task 16 of 20 in Humanity's Last Exam
Humanity's Last Exam item 40
Muse Wonder's two runs came closest to the time asked: 15 s when asked for 15 s, and 3 min 5 s when asked for 4 min. MiMo UltraSpeed worked about the same whatever it was asked: 42 s when asked for 1 min 15 s, and 38 s when asked for 20 min.
- Asked for
- 8 s to 24 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 | 30 s | 1 min 15 s | 1 min 14 s 9.89× longer asked 8 s | 3 min 49 s 1.91× longer asked 2 min |
| Argon | 30 s | 1 min 11 s | 2 min 31 s 20.19× longer asked 8 s | 2 min 22 s 1.18× longer asked 2 min |
| Sol 6.1 | 1 min 30 s | 44 s | 1 min 3 s 2.78× longer asked 23 s | 6 min 13 s Within 5% asked 6 min |
| MiMo Pro | 6 min | 3 min 18 s | 5 min 10 s 3.45× longer asked 1 min 30 s | 21 min 11 s 1.13× shorter asked 24 min |
| MiMo UltraSpeed | 5 min | 16 s | 42 s 1.80× shorter asked 1 min 15 s | 38 s 31.50× shorter asked 20 min |
| Muse Wonder | 1 min | 11 s | 15 s Within 5% asked 15 s | 3 min 5 s 1.30× shorter asked 4 min |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Phan et al., 2025