AppWorld, AgentTime v1.1, asked for four times its own forecast
Claude Opus 5.5 on “Turn a note-based meeting schedule into correctly timed reminder emails”
Shorter
8.64× shorterStopped at 12% of the time asked, 24 min 46 s short.
- Asked for
- 28 minutes
- Four times its own forecast of 7 min.
- Worked
- 3 min 14 s
- Grade
- Not graded
- How it ended
- Ended on its own
Every detail of this run
- Agent
- Claude Opus 5.5, in Claude Code
- Benchmark
- AppWorld
- Asked for
- 28 minutes, four times its own forecast of 7 min: “Please work on this task for a full 28 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 3 min 14 s (194 s)
- Ratio
- 0.12× the request
- How it ended
- Ended on its own
- Grade
- Not graded
- Clock tool
- The agent could call a wall-clock tool to read the current time.
- Wording
- The paper's standard sentence, sent in AgentTime v1.1.
- Run id
- 4ezoicybqo
- Data release
- 2 Nov 2026
Related runs
| Opus 5.5 on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Quarter | 1.75 minutes | 1 min 25 s | 1.24× shorter | |
| Four times (this run) | 28 minutes | 3 min 14 s | 8.64× shorter | |
| Other agents, four times their own forecast | ||||
| Gemini 4 Argon | 480 seconds | 8 min 27 s | 1.06× longer | |
| GPT 6.1 Sol | 18 minutes | 18 min 26 s | Within 5% | |
| MiMo V2.6 Pro | 140 minutes | 4 min 12 s | 33.33× shorter | |
| MiMo V2.6 Pro UltraSpeed | 100 minutes | 3 min 4 s | 32.68× shorter | |
| Muse Wonder 1.0 | 56 minutes | 56 s | 60.22× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Trivedi et al., 2024 (ACL)