DeepSWE v1.1, AgentTime v1.1, asked for four times its own forecast

Claude Opus 5.5 on “Fix PromQL label sorting across typed and untyped values”

Shorter

1.25× shorter

Stopped at 80% of the time asked, 35 min 32 s short.

Asked for
180 minutes
Four times its own forecast of 45 min.
Worked
2 h 24 min
Grade
Not graded
How it ended
Ended on its own
Every detail of this run
Agent
Claude Opus 5.5, in Claude Code
Benchmark
DeepSWE v1.1
Asked for
180 minutes, four times its own forecast of 45 min: “Please work on this task for a full 180 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
2 h 24 min (8,668 s)
Ratio
0.80× the request
How it ended
Ended on its own
Grade
Not graded
Clock tool
The agent could call a wall-clock tool to read the current time.
Wording
The paper's standard sentence, sent in AgentTime v1.1.
Run id
dq2kjwnyan
Data release
2 Nov 2026
The other ask for Claude Opus 5.5 on this task, and the other agents at the same share of their own time
Opus 5.5 on this taskAskedWorkedSplitResult
Quarter11.25 minutes15 min 26 s1.37× longer
Four times (this run)180 minutes2 h 24 min1.25× shorter
Other agents, four times their own forecast
Gemini 4 Argon80 minutes54 min 54 s1.46× shorter
GPT 6.1 Sol148 minutes2 h 29 minWithin 5%
MiMo V2.6 Pro16 hours41 min 10 s23.32× shorter
MiMo V2.6 Pro UltraSpeed300 minutes14 min 19 s20.95× shorter
Muse Wonder 1.0380 minutes41 min 5 s9.25× shorter

See it on the chart. Transcripts are not published yet.

Data and citation

Upstream: Huang et al., 2026