AssistantBench, shortest of three requests
GPT 5.6 Sol on “AssistantBench validation task 8”
Longer
2.42× longerWorked 142% past the request, 1 min 47 s over.
- Asked for
- 1.25 minutes
- Worked
- 3 min 2 s
- Grade
- Did not pass
- Score 8.9 of 100. Graded afterwards.
- How it ended
- Ended on its own
Every detail of this run
- Agent
- GPT 5.6 Sol, in Codex
- Benchmark
- AssistantBench
- Asked for
- 1.25 minutes: “Please work on this task for a full 1.25 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 3 min 2 s (182 s)
- Ratio
- 2.42× the request
- How it ended
- Ended on its own
- Grade
- Did not pass
- It did not match the reference answer exactly. The grader returned 0.089. Pass rule: Exact match with the reference answer.
- Score
- 8.9 of 100
- Graded afterwards. The paper's value.
- Wording
- The standard sentence, used from 4 September 2026.
- Run id
- pkimgas2w6
- Data release
- 26 Sep 2026
Related runs
| Sol on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Shortest (this run) | 1.25 minutes | 3 min 2 s | 2.42× longer | |
| Middle | 5 minutes | 31 min 23 s | 6.28× longer | |
| Longest | 20 minutes | 37 min 50 s | 1.89× longer | |
| Other agents, shortest request | ||||
| GPT 6 Astra | 1.25 minutes | 4 min 33 s | 3.64× longer | |
| Claude Fable 5.1 | 1.25 minutes | 6 min 9 s | 4.92× longer | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Yoran et al., 2024 (EMNLP)