Task 13 of 20 in Humanity's Last Exam
Humanity's Last Exam item 37
Sol 6.1's two runs came closest to the time asked: 53 s when asked for 15 s, and 4 min 12 s when asked for 4 min. Argon's two runs landed furthest: 7 min 31 s when asked for 6 s, and 1 min 51 s when asked for 1 min 36 s.
- Asked for
- 6 s to 20 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 | 1 min 30 s | 4 min 34 s | 1 min 42 s 4.52× longer asked 23 s | 6 min 57 s 1.16× longer asked 6 min |
| Argon | 24 s | 9 min 2 s | 7 min 31 s 75.15× longer asked 6 s | 1 min 51 s 1.16× longer asked 1 min 36 s |
| Sol 6.1 | 1 min | 4 min 3 s | 53 s 3.55× longer asked 15 s | 4 min 12 s Within 5% asked 4 min |
| MiMo Pro | 1 min | 1 min 20 s | 3 min 8 s 12.55× longer asked 15 s | 11 min 17 s 2.82× longer asked 4 min |
| MiMo UltraSpeed | 5 min | 20 s | 1 min 8 s 1.11× shorter asked 1 min 15 s | 1 min 9 s 17.34× shorter asked 20 min |
| Muse Wonder | 1 min | 37 s | 43 s 2.85× longer asked 15 s | 1 min 36 s 2.51× 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