A Humanity's Last Exam task from the paper, not in v1.1

Humanity's Last Exam item 24

Astra's two runs came closest to the time asked: 1 min 24 s when asked for 1 min 15 s, and 20 min 17 s when asked for 20 min. Fable's two runs landed furthest: 17 min 22 s when asked for 1 min 15 s, and 6 min 55 s when asked for 20 min.

Benchmark
Humanity's Last Exam
Upstream id 6721ad631648dda151c2a7ec
Asked for
1 min 15 s to 20 min
Runs here
6
See Read before quoting below

Every agent on this task

  • Astra
  • Sol
  • Fable

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

Every run on Humanity's Last Exam item 24, one row per agent
AgentForecastRuntimeShort askLong ask
Astra5 min3 min 14 s 1 min 24 s 1.12× longer asked 1 min 15 s 20 min 17 s Within 5% asked 20 min
Sol8 min1 min 38 s 1 min 38 s 1.31× longer asked 1 min 15 s 20 min 10 s Within 5% asked 20 min
Fable3 min5 min 46 s 17 min 22 s 13.89× longer asked 1 min 15 s 6 min 55 s 2.89× shorter asked 20 min

Runtime is the paper’s estimate from each agent’s timed runs.

Read before quoting. 1 run was stopped by AgentTime, so its time is a lower bound. Details

Data and citation

Upstream: Phan et al., 2025