OSWorld 2.0, AgentTime v1.1, asked for a quarter of its own forecast
Muse Wonder 1.0 on “Twenty recurring meetings”
Shorter
1.54× shorterStopped at 65% of the time asked, 10 min 31 s short.
- Asked for
- 30 minutes
- A quarter of its own forecast of 2 h.
- Worked
- 19 min 29 s
- Grade
- Not graded
- How it ended
- Ended on its own
Every detail of this run
- Agent
- Muse Wonder 1.0, in Muse Code
- Benchmark
- OSWorld 2.0
- Asked for
- 30 minutes, a quarter of its own forecast of 2 h: “Please work on this task for a full 30 minutes. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 19 min 29 s (1,169 s)
- Ratio
- 0.65× 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
- 7tsrq7bidt
- Data release
- 2 Nov 2026
Related runs
| Muse Wonder on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Quarter (this run) | 30 minutes | 19 min 29 s | 1.54× shorter | |
| Four times | 480 minutes | no run | ||
| Other agents, a quarter of their own forecast | ||||
| Claude Opus 5.5 | 35 minutes | 46 min 46 s | 1.34× longer | |
| Gemini 4 Argon | 7 minutes | 28 min 42 s | 4.10× longer | |
| GPT 6.1 Sol | 22.5 minutes | 22 min 46 s | Within 5% | |
| MiMo V2.6 Pro | 17.5 minutes | 4 h 11 min | 14.33× longer | |
| MiMo V2.6 Pro UltraSpeed | 18 minutes | 19 min 56 s | 1.11× longer | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Yuan et al., 2026