Agents' Last Exam, AgentTime v1.1, asked for four times its own forecast

Muse Wonder 1.0 on “Apple FY2024 Balance Sheet Reconstruction”

Shorter

29.09× shorter

Stopped at 3% of the time asked, 30 min 54 s short.

Asked for
32 minutes
Four times its own forecast of 8 min.
Worked
1 min 6 s
Grade
Not graded
How it ended
Ended on its own
Every detail of this run
Agent
Muse Wonder 1.0, in Muse Code
Asked for
32 minutes, four times its own forecast of 8 min: “Please work on this task for a full 32 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
1 min 6 s (66 s)
Ratio
0.03× 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
33zs5xsrwd
Data release
2 Nov 2026
The other ask for Muse Wonder 1.0 on this task, and the other agents at the same share of their own time
Muse Wonder on this taskAskedWorkedSplitResult
Quarter2 minutes43 s2.78× shorter
Four times (this run)32 minutes1 min 6 s29.09× shorter
Other agents, four times their own forecast
Claude Opus 5.524 minutes9 min 8 s2.63× shorter
Gemini 4 Argon480 seconds3 min 6 s2.58× shorter
GPT 6.1 Sol600 seconds10 min 23 sWithin 5%
MiMo V2.6 Pro48 minutes4 min 43 s10.19× shorter
MiMo V2.6 Pro UltraSpeed40 minutes2 min 30 s16.00× shorter

See it on the chart. Transcripts are not published yet.

Data and citation

Upstream: Sun et al., 2026