Task 7 of 20 in Humanity's Last Exam
Humanity's Last Exam item 31
Muse Wonder's two runs came closest to the time asked: 1 min 48 s when asked for 2 min 15 s, and 14 min 28 s when asked for 36 min. MiMo UltraSpeed worked about the same whatever it was asked: 5 min 44 s when asked for 7 min 30 s, and 5 min 55 s when asked for 2 h.
- Asked for
- 53 s 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 | 13 min | 32 min 42 s | 18 min 14 s 5.61× longer asked 3 min 15 s | 50 min 55 s Within 5% asked 52 min |
| Argon | 3 min 30 s | 21 min 1 s | 1 min 10 s 1.33× longer asked 53 s | 46 min 42 s 3.34× longer asked 14 min |
| Sol 6.1 | 8 min | 21 min 55 s | 9 min 57 s 4.98× longer asked 2 min | 32 min 12 s Within 5% asked 32 min |
| MiMo Pro | 45 min | 1 min 48 s | 1 h 10 min 6.23× longer asked 11 min 15 s | 1 h 18 min 2.31× shorter asked 3 h |
| MiMo UltraSpeed | 30 min | 8 s | 5 min 44 s 1.31× shorter asked 7 min 30 s | 5 min 55 s 20.26× shorter asked 2 h |
| Muse Wonder | 9 min | 32 min 40 s | 1 min 48 s 1.25× shorter asked 2 min 15 s | 14 min 28 s 2.49× shorter asked 36 min |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Phan et al., 2025