Task 18 of 20 in Humanity's Last Exam
Humanity's Last Exam item 42
Sol 6.1's two runs came closest to the time asked: 43 s when asked for 30 s, and 8 min 10 s when asked for 8 min. MiMo UltraSpeed worked about the same whatever it was asked: 2 min 40 s when asked for 1 min, and 2 min 29 s when asked for 16 min.
- Asked for
- 11 s to 16 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 | 3 min 22 s | 35 s 2.35× longer asked 15 s | 5 min 1 s 1.26× longer asked 4 min |
| Argon | 42 s | 3 min 30 s | 36 s 3.44× longer asked 11 s | 3 min 6 s 1.11× longer asked 2 min 48 s |
| Sol 6.1 | 2 min | 2 min 24 s | 43 s 1.43× longer asked 30 s | 8 min 10 s Within 5% asked 8 min |
| MiMo Pro | 4 min | 3 min 8 s | 5 min 6 s 5.10× longer asked 1 min | 31 min 21 s 1.96× longer asked 16 min |
| MiMo UltraSpeed | 4 min | 27 s | 2 min 40 s 2.66× longer asked 1 min | 2 min 29 s 6.43× shorter asked 16 min |
| Muse Wonder | 2 min | 37 s | 50 s 1.65× longer asked 30 s | 4 min 14 s 1.89× shorter asked 8 min |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Phan et al., 2025