AssistantBench, longest of three requests
GPT 6 Astra on “AssistantBench validation task 16”
Shorter
4.12× shorterStopped at 24% of the time asked, 15 min 8 s short.
- Asked for
- 20 minutes
- Worked
- 4 min 52 s
- Grade
- Did not pass
- Score 0 of 100. Graded afterwards.
- How it ended
- Ended on its own
Every detail of this run
- Agent
- GPT 6 Astra, in Codex
- Benchmark
- AssistantBench
- Asked for
- 20 minutes: “Please work on this task for a full 20 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 4 min 52 s (292 s)
- Ratio
- 0.24× 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. Pass rule: Exact match with the reference answer.
- Score
- 0 of 100
- Graded afterwards. The paper's value.
- Wording
- The standard sentence, used from 4 September 2026.
- Run id
- 5alox7l6wk
- Data release
- 26 Sep 2026
Related runs
| Astra on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Shortest | 1.25 minutes | 2 min 58 s | 2.38× longer | |
| Middle | 5 minutes | 5 min 47 s | 1.16× longer | |
| Longest (this run) | 20 minutes | 4 min 52 s | 4.12× shorter | |
| Other agents, longest request | ||||
| GPT 5.6 Sol | 20 minutes | 19 min 20 s | Within 5% | |
| Claude Fable 5.1 | 20 minutes | 4 min 22 s | 4.58× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Yoran et al., 2024 (EMNLP)