Task 2 of 4 in PaperBench
FRE
Astra's two runs followed the time asked: 10 min 9 s when asked for 10 min, and 2 h 31 min when asked for 2 h 30 min. MiMo Pro's two runs landed furthest: 1 h 33 min when asked for 4 h 26 min, and 18 h 50 min when asked for 71 h 6 min.
- Benchmark
- PaperBench
- Asked for
- 10 min to 97 h 36 min
- Runs here
- 16
- 8 agents
Every agent on this task
- Astra
- Sol
- Fable
- Opus 5.5
- Argon
- Sol 6.1
- MiMo Pro
- MiMo UltraSpeed
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 | 24 h | 26 min 57 s | 10 min 9 s Within 5% asked 10 min | 2 h 31 min Within 5% asked 2 h 30 min |
| Sol | 12 h | 12 min 50 s | 6 min 27 s 1.55× shorter asked 10 min | 2 h 31 min Within 5% asked 2 h 30 min |
| Fable | 12 h | 14 min 3 s | 5 min 1 s 1.99× shorter asked 10 min | 42 min 3 s 3.57× shorter asked 2 h 30 min |
| Opus 5.5 | 18 h 20 min | 33 h 27 min | 5 h 36 min 1.09× shorter asked 6 h 6 min | 38 h 3 min 2.56× shorter asked 97 h 36 min |
| Argon | 10 h 20 min | 2 h 31 min | 1 h 8 min 1.79× longer asked 37 min 54 s | 7 h 22 min 1.37× shorter asked 10 h 6 min |
| Sol 6.1 | 240 h | 2 h 20 min | 35 min 32 s Within 5% asked 35 min | 18 h 38 min 2.00× longer asked 9 h 20 min |
| MiMo Pro | 216 h | 17 h 47 min | 1 h 33 min 2.86× shorter asked 4 h 26 min | 18 h 50 min 3.78× shorter asked 71 h 6 min |
| MiMo UltraSpeed | 96 h | 6 h 16 min | 25 min 50 s 3.65× shorter asked 1 h 34 min | 17 h 9 min 1.46× shorter asked 25 h 6 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: Starace et al., 2025