PaperBench, longest of three requests
GPT 5.6 Sol on “BAM”
Longer
1.14× longerWorked 14% past the request, 11 min 10 s over.
- Asked for
- 80 minutes
- Worked
- 1 h 31 min
- Grade
- Score 62.8 of 100
- Judged afterwards.
- How it ended
- Ended on its own
Every detail of this run
- Agent
- GPT 5.6 Sol, in Codex
- Benchmark
- PaperBench
- Task
- BAM
- Asked for
- 80 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.”
- Worked
- 1 h 31 min (5,470 s)
- Ratio
- 1.14× the request
- How it ended
- Ended on its own
- Grade
- 0.628
- Replication score: a weighted rubric of pass/fail leaf checks, 0 to 1, after a full reproduction attempt. No official pass rule.
- Score
- 62.8 of 100
- Judged afterwards. The paper's value.
- Wording
- The standard sentence, used from 4 September 2026.
- Run id
- 6ymdpocc4u
- Data release
- 26 Sep 2026
Related runs
| Sol on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Shortest | 5 minutes | 8 min 39 s | 1.73× longer | |
| Middle | 20 minutes | 24 min 7 s | 1.21× longer | |
| Longest (this run) | 80 minutes | 1 h 31 min | 1.14× longer | |
| Other agents, longest request | ||||
| GPT 6 Astra | 80 minutes | 1 h 21 min | Within 5% | |
| Claude Fable 5.1 | 80 minutes | 22 min 1 s | 3.63× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Starace et al., 2025