Task 10 of 20 in Humanity's Last Exam
Humanity's Last Exam item 34
Sol 6.1's two runs came closest to the time asked: 3 min 20 s when asked for 2 min, and 32 min 49 s when asked for 32 min. MiMo UltraSpeed worked about the same whatever it was asked: 9 min 54 s when asked for 11 min 15 s, and 5 min 18 s when asked for 3 h.
- Asked for
- 38 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 | 10 min | 15 min 2 s | 8 min 15 s 3.30× longer asked 2 min 30 s | 43 min 37 s 1.09× longer asked 40 min |
| Argon | 2 min 30 s | 4 min 4 s | 4 min 35 s 7.33× longer asked 38 s | 10 min 46 s 1.08× longer asked 10 min |
| Sol 6.1 | 8 min | 5 min 3 s | 3 min 20 s 1.67× longer asked 2 min | 32 min 49 s Within 5% asked 32 min |
| MiMo Pro | 38 min | 44 min 24 s | 30 min 44 s 3.24× longer asked 9 min 30 s | 1 h 8 min 2.23× shorter asked 2 h 32 min |
| MiMo UltraSpeed | 45 min | 4 min 26 s | 9 min 54 s 1.14× shorter asked 11 min 15 s | 5 min 18 s 33.92× shorter asked 3 h |
| Muse Wonder | 5 min | 10 min 55 s | 8 min 10 s 6.53× longer asked 1 min 15 s | 5 min 39 s 3.54× shorter asked 20 min |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Phan et al., 2025