PaperBench, AgentTime v1.1, asked for four times its runtime with no time asked

Claude Opus 5.5 on “Challenges in Training PINNs: A Loss Landscape Perspective”

Shorter

3.40× shorter

Stopped at 29% of the time asked, 39 h 5 min short.

Asked for
55.4 hours
Four times its runtime with no time asked (13 h 51 min).
Worked
16 h 19 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
PaperBench
Asked for
55.4 hours, four times its runtime with no time asked (13 h 51 min): “Please work on this task for a full 55.4 hours. Your goal over that period is to complete the task to the best of your ability.”
Worked
16 h 19 min (58,722 s)
Ratio
0.29× 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
7iz5o2ksnb
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
Quarter3.46 hours2 h 35 min1.34× shorter
Four times (this run)55.4 hours16 h 19 min3.40× shorter
Other agents, four times their own runtime
Gemini 4 Argon253 minutes2 h 21 min1.80× shorter
GPT 6.1 Sol442 minutes7 h 22 minWithin 5%
MiMo V2.6 Pro UltraSpeed35 hours12 h 53 min2.72× shorter

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

Data and citation

Upstream: Starace et al., 2025