Task 12 of 12 in CORE-Bench
Human Activities Recognition on PAMAP2 dataset using Random Forest
Sol 6.1's two runs followed the time asked: 11 min 35 s when asked for 11 min 15 s, and 3 h 1 min when asked for 3 h. MiMo Pro worked about the same whatever it was asked: 2 min 56 s when asked for 15 min, and 3 min 49 s when asked for 4 h.
- Asked for
- 2 min to 4 h 40 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 | 20 min | 16 min 4 s | 12 min 3 s 2.41× longer asked 5 min | 53 min 20 s 1.50× shorter asked 1 h 20 min |
| Argon | 8 min | 14 min 7 s | 3 min 25 s 1.71× longer asked 2 min | 32 min 49 s Within 5% asked 32 min |
| Sol 6.1 | 45 min | 2 min 50 s | 11 min 35 s Within 5% asked 11 min 15 s | 3 h 1 min Within 5% asked 3 h |
| MiMo Pro | 1 h | 10 min 23 s | 2 min 56 s 5.12× shorter asked 15 min | 3 min 49 s 62.99× shorter asked 4 h |
| MiMo UltraSpeed | 1 h 10 min | 4 min 28 s | 9 min 4 s 1.93× shorter asked 17 min 30 s | 9 min 47 s 28.63× shorter asked 4 h 40 min |
| Muse Wonder | 35 min | 8 min 32 s | 2 min 36 s 3.37× shorter asked 8 min 45 s | 31 min 2 s 4.51× shorter asked 2 h 20 min |
No runs match these filters.
Runtime is each agent’s run with no time asked.
Data and citation
Upstream: Nadgir et al., 2026