AppWorld, middle of three requests
Claude Fable 5.1 on “Execute every task in a camping project across shopping, expenses, payments, invitations, and task closeout”
Shorter
1.66× shorterStopped at 60% of the time asked, 3 min 58 s short.
- Asked for
- 10 minutes
- Worked
- 6 min 2 s
- Grade
- Did not pass
- Score 0 of 100. Graded afterwards.
- How it ended
- Ended on its own
Every detail of this run
- Agent
- Claude Fable 5.1, in Claude Code
- Benchmark
- AppWorld
- Asked for
- 10 minutes: “Please work on this task for a full 10 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 6 min 2 s (362 s)
- Ratio
- 0.60× the request
- How it ended
- Ended on its own
- Grade
- Did not pass
- It did not pass every final-state test. The grader returned 0. Pass rule: Every final-state test passes.
- Score
- 0 of 100
- Graded afterwards. The paper's value.
- Wording
- The standard sentence, used from 4 September 2026.
- Run id
- 657xfgfg5j
- Data release
- 26 Sep 2026
Related runs
| Fable on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Shortest | 2.5 minutes | 6 min 51 s | 2.74× longer | |
| Middle (this run) | 10 minutes | 6 min 2 s | 1.66× shorter | |
| Longest | 40 minutes | 7 min 31 s | 5.32× shorter | |
| Other agents, middle request | ||||
| GPT 6 Astra | 10 minutes | 10 min 38 s | 1.06× longer | |
| GPT 5.6 Sol | 10 minutes | 10 min 42 s | 1.07× longer | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Trivedi et al., 2024 (ACL)