ProgramBench, longest of three requests
Claude Fable 5.1 on “tomarrell/wrapcheck”
Longer
2.00× longerWorked 100% past the request, 4 h 10 min over.
- Asked for
- 250 minutes
- Worked
- 8 h 20 min
- Grade
- Did not pass
- Score 80.4 of 100. Graded afterwards.
- How it ended
- Stopped by the safety cutoff
Every detail of this run
- Agent
- Claude Fable 5.1, in Claude Code
- Benchmark
- ProgramBench
- Asked for
- 250 minutes: “Please work on this task for a full 250 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 8 h 20 min (30,000 s)
- Ratio
- 2.00× the request
- How it ended
- Stopped by the safety cutoff
- Grade
- Did not pass
- It did not pass every hidden test. The grader returned 0.804. Pass rule: Resolved: every hidden test passes.
- Score
- 80.4 of 100
- Graded afterwards. The paper's value.
- Wording
- The standard sentence, used from 4 September 2026.
- Run id
- b2hguakdt2
- Data release
- 26 Sep 2026
Related runs
| Fable on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Shortest | 15 minutes | 1 h 46 min | 7.04× longer | |
| Middle | 60 minutes | 8 h 20 min | 8.33× longer | |
| Longest (this run) | 250 minutes | 8 h 20 min | 2.00× longer | |
| Other agents, longest request | ||||
| GPT 6 Astra | 250 minutes | 4 h 10 min | Within 5% | |
| GPT 5.6 Sol | 250 minutes | 4 h 11 min | Within 5% | |
See it on the chart. Transcripts are not published yet.
Read before quoting. This run was stopped by the safety cutoff, so its time is a lower bound: the agent might have kept going. Details
Data and citation
Upstream: Yang et al., 2026