Task 8 of 12 in GPQA Diamond
GPQA Diamond closed-book question 16 (Physics)
Astra's two runs came closest to the time asked: 1 min 26 s when asked for 1 min 15 s, and 20 min 14 s when asked for 20 min. MiMo UltraSpeed's two runs landed furthest: 10 s when asked for 30 s, and 26 s when asked for 8 min.
- Asked for
- 5 s to 20 min
- Runs here
- 18
- 9 agents
Every agent on this task
- Astra
- Sol
- Fable
- 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 |
|---|---|---|---|---|
| Astra | 1 min | 2 min 9 s | 1 min 26 s 1.14× longer asked 1 min 15 s | 20 min 14 s Within 5% asked 20 min |
| Sol | 1 min | 36 s | 1 min 27 s 1.16× longer asked 1 min 15 s | 20 min 20 s Within 5% asked 20 min |
| Fable | 1 min | 1 min 17 s | 1 min 37 s 1.30× longer asked 1 min 15 s | 1 min 10 s 17.14× shorter asked 20 min |
| Opus 5.5 | 24 s | 17 s | 24 s 3.93× longer asked 6 s | 1 min 53 s 1.18× longer asked 1 min 36 s |
| Argon | 18 s | 31 s | 37 s 8.18× longer asked 5 s | 1 min 23 s 1.16× longer asked 1 min 12 s |
| Sol 6.1 | 45 s | 11 s | 19 s 1.64× longer asked 11 s | 3 min 11 s 1.06× longer asked 3 min |
| MiMo Pro | 3 min | 43 s | 30 s 1.48× shorter asked 45 s | 1 min 46 s 6.79× shorter asked 12 min |
| MiMo UltraSpeed | 2 min | 7 s | 10 s 2.88× shorter asked 30 s | 26 s 18.18× shorter asked 8 min |
| Muse Wonder | 1 min | 6 s | 27 s 1.77× longer asked 15 s | 1 min 4 s 3.74× shorter asked 4 min |
No runs match these filters.
Runtime is a v1.1 agent’s run with no time asked, and for the paper’s three agents the paper’s estimate from their timed runs.
Data and citation
Upstream: Rein et al., 2024 (COLM)