Task 19 of 20 in Humanity's Last Exam
Humanity's Last Exam item 43
Sol 6.1's two runs came closest to the time asked: 54 s when asked for 18 s, and 4 min 58 s when asked for 4 min 48 s. MiMo Pro's two runs landed furthest: 8 min 47 s when asked for 15 s, and 2 min 16 s when asked for 4 min.
- Asked for
- 12 s to 12 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 24 s | 2 min 33 s 10.17× longer asked 15 s | 8 min 36 s 2.15× longer asked 4 min |
| Argon | 48 s | 7 min 42 s | 1 min 30 s 7.50× longer asked 12 s | 3 min 26 s 1.07× longer asked 3 min 12 s |
| Sol 6.1 | 1 min 12 s | 2 min 55 s | 54 s 3.02× longer asked 18 s | 4 min 58 s Within 5% asked 4 min 48 s |
| MiMo Pro | 1 min | 5 min 1 s | 8 min 47 s 35.15× longer asked 15 s | 2 min 16 s 1.76× shorter asked 4 min |
| MiMo UltraSpeed | 3 min | 33 s | 44 s Within 5% asked 45 s | 1 min 6 s 10.89× shorter asked 12 min |
| Muse Wonder | 1 min | 16 s | 28 s 1.89× longer asked 15 s | 1 min 38 s 2.45× 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