Humanity's Last Exam, AgentTime v1.1, asked for four times its own forecast
Muse Wonder 1.0 on “Humanity's Last Exam item 35”
Shorter
1.98× shorterStopped at 51% of the time asked, 59 s short.
- Asked for
- 120 seconds
- Four times its own forecast of 30 s.
- Worked
- 1 min 1 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
- Humanity's Last Exam
- Asked for
- 120 seconds, four times its own forecast of 30 s: “Please work on this task for a full 120 seconds. Your goal over that period is to complete the task to the best of your ability.”
- Worked
- 1 min 1 s (61 s)
- Ratio
- 0.51× 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
- dmzyxzpwyd
- Data release
- 2 Nov 2026
Related runs
| Muse Wonder on this task | Asked | Worked | Split | Result |
|---|---|---|---|---|
| Quarter | 7.5 seconds | 17 s | 2.25× longer | |
| Four times (this run) | 120 seconds | 1 min 1 s | 1.98× shorter | |
| Other agents, four times their own forecast | ||||
| Claude Opus 5.5 | 240 seconds | 4 min 44 s | 1.18× longer | |
| Gemini 4 Argon | 120 seconds | 2 min 23 s | 1.19× longer | |
| GPT 6.1 Sol | 120 seconds | 2 min 9 s | 1.07× longer | |
| MiMo V2.6 Pro | 480 seconds | 1 min 47 s | 4.49× shorter | |
| MiMo V2.6 Pro UltraSpeed | 20 minutes | 22 s | 55.05× shorter | |
See it on the chart. Transcripts are not published yet.
Data and citation
Upstream: Phan et al., 2025