Agents' Last Exam, longest of three requests
GPT 5.6 Sol on “Route a KiCad PCB from the staged schematic”
Shorter
1.63× shorterStopped at 61% of the time asked, 57 min 45 s short.
- Asked for
- 150 minutes
- Worked
- 1 h 32 min
- Grade
- Did not pass
- Score 0 of 100. Graded afterwards.
- How it ended
- Ended on its own
Every detail of this run
- Agent
- GPT 5.6 Sol, in Codex
- Benchmark
- Agents' Last Exam
- Asked for
- 150 minutes: “Please work on this task for a full 150 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 1 h 32 min (5,535 s)
- Ratio
- 0.61× the request
- How it ended
- Ended on its own
- Grade
- Did not pass
- It did not pass the task's checker in full. The grader returned 0. Pass rule: Full pass: a checker score of 1.
- Score
- 0 of 100
- Graded afterwards. The paper's value.
- Wording
- The standard sentence, used from 4 September 2026.
- Run id
- wyqxvvvwyj
- Data release
- 26 Sep 2026
Related runs
| Sol on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Shortest | 10 minutes | 1 h 57 min | 11.70× longer | |
| Middle | 40 minutes | 1 h 35 min | 2.38× longer | |
| Longest (this run) | 150 minutes | 1 h 32 min | 1.63× shorter | |
| Other agents, longest request | ||||
| GPT 6 Astra | 150 minutes | 2 h 32 min | Within 5% | |
| Claude Fable 5.1 | 150 minutes | 39 min 36 s | 3.79× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Sun et al., 2026