ProgramBench, AgentTime v1.1, asked for four times its own forecast
Muse Wonder 1.0 on “facebook/zstd”
Shorter
464.16× shorterStopped at 0% of the time asked, 67 h 51 min short.
- Asked for
- 68 hours
- Four times its own forecast of 17 h.
- Worked
- 8 min 47 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
- ProgramBench
- Task
- facebook/zstd
- Asked for
- 68 hours, four times its own forecast of 17 h: “Please work on this task for a full 68 hours. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 8 min 47 s (527 s)
- Ratio
- 0.00× 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
- 7wjuzyngmj
- Data release
- 2 Nov 2026
Related runs
| Muse Wonder on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Quarter | 4.25 hours | 5 min 29 s | 46.53× shorter | |
| Four times (this run) | 68 hours | 8 min 47 s | 464.16× shorter | |
| Other agents, four times their own forecast | ||||
| Claude Opus 5.5 | 720 minutes | 11 h 49 min | Within 5% | |
| Gemini 4 Argon | 300 minutes | 41 min 40 s | 7.20× shorter | |
| GPT 6.1 Sol | 480 minutes | 8 h 1 min | Within 5% | |
| MiMo V2.6 Pro | 160 hours | 2 h 11 min | 73.09× shorter | |
| MiMo V2.6 Pro UltraSpeed | 140 hours | 56 min 41 s | 148.17× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Yang et al., 2026