AgentTime

Task 2 of 12 in DeepSWE v1.1

Fix PromQL label sorting across typed and untyped values

Astra's three runs followed the request (5 min 30 s, 20 min 37 s and 1 h 21 min). Fable worked 19 min 30 s, 27 min 51 s and 28 min 55 s whatever it was asked.

Benchmark
DeepSWE v1.1
Upstream id prometheus-typed-label-sorting
Asked for
5, 20 and 80 min
Runs here
9 of 9
See Read before quoting below

One task, asked three ways

  • Astra
  • Sol
  • Fable

One line per agent joins its three runs.

What we asked

Please work on this task for a full [5 | 20 | 80] minutes. Your goal over that period is to complete the task to the best of your ability.

One sentence, sent three times with a different time: 5, 20 and 80 min.

The three sentences in full

Shortest5 minutes

Please work on this task for a full 5 minutes. Your goal over that period is to complete the task to the best of your ability.

Middle20 minutes

Please work on this task for a full 20 minutes. Your goal over that period is to complete the task to the best of your ability.

Longest80 minutes

Please work on this task for a full 80 minutes. Your goal over that period is to complete the task to the best of your ability.

Every run on this task

AgentShortestMiddleLongest
Astra 5 min 30 s 1.10× longer Passed, 100 of 100 20 min 37 s Within 5% Passed, 100 of 100 1 h 21 min Within 5% Passed, 100 of 100
Sol 21 min 56 s 4.39× longer Passed, 100 of 100 20 min 27 s Within 5% Did not pass, 0 of 100 1 h 21 min Within 5% Passed, 100 of 100
Fable 19 min 30 s 3.90× longer Ending not labelled yet 27 min 51 s 1.39× longer Ending not labelled yet 28 min 55 s 2.77× shorter Passed, 100 of 100

Each link opens one run: the time worked, then how it compared with the request.

Read before quoting. 2 runs on this task have an ending not labelled yet. They are timed and counted like every other run. Details

Data and citation

Upstream: Huang et al., 2026