Terminal-Bench 4.0, AgentTime v1.1, asked for a quarter of its own forecast
Muse Wonder 1.0 on “mp-checkpoint-consolidation”
Shorter
1.39× shorterStopped at 72% of the time asked, 5 min 38 s short.
- Asked for
- 20 minutes
- A quarter of its own forecast of 1 h 20 min.
- Worked
- 14 min 22 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
- Terminal-Bench 4.0
- Asked for
- 20 minutes, a quarter of its own forecast of 1 h 20 min: “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
- 14 min 22 s (862 s)
- Ratio
- 0.72× 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
- wn45lb3zhh
- Data release
- 2 Nov 2026
Related runs
| Muse Wonder on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Quarter (this run) | 20 minutes | 14 min 22 s | 1.39× shorter | |
| Four times | 320 minutes | 33 min 19 s | 9.60× shorter | |
| Other agents, a quarter of their own forecast | ||||
| Claude Opus 5.5 | 6.25 minutes | 2 min 11 s | 2.85× shorter | |
| Gemini 4 Argon | 3 minutes | 22 min 21 s | 7.45× longer | |
| GPT 6.1 Sol | 4.5 minutes | 9 min 2 s | 2.01× longer | |
| MiMo V2.6 Pro | 26.25 minutes | 58 min 7 s | 2.21× longer | |
| MiMo V2.6 Pro UltraSpeed | 1 hour | 41 min 19 s | 1.45× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Merrill et al., 2026 (ICLR)