PaperBench, AgentTime v1.1, asked for four times its runtime with no time asked
Claude Opus 5.5 on “FRE”
Shorter
2.56× shorterStopped at 39% of the time asked, 59 h 33 min short.
- Asked for
- 97.6 hours
- Four times its runtime with no time asked (24 h 25 min).
- Worked
- 38 h 3 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
- Task
- FRE
- Asked for
- 97.6 hours, four times its runtime with no time asked (24 h 25 min): “Please work on this task for a full 97.6 hours. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 38 h 3 min (136,988 s)
- Ratio
- 0.39× 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
- afxzkcautb
- Data release
- 2 Nov 2026
Related runs
| Opus 5.5 on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Quarter | 6.1 hours | 5 h 36 min | 1.09× shorter | |
| Four times (this run) | 97.6 hours | 38 h 3 min | 2.56× shorter | |
| Other agents, four times their own runtime | ||||
| Gemini 4 Argon | 606 minutes | 7 h 22 min | 1.37× shorter | |
| GPT 6.1 Sol | 560 minutes | 18 h 38 min | 2.00× longer | |
| MiMo V2.6 Pro | 71.1 hours | 18 h 50 min | 3.78× shorter | |
| MiMo V2.6 Pro UltraSpeed | 25.1 hours | 17 h 9 min | 1.46× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Starace et al., 2025