OSWorld 2.0, AgentTime v1.1, asked for a quarter of its own forecast

Muse Wonder 1.0 on “Local event booking QA”

Longer

11.77× longer

Worked 11.8 times as long as asked, 1 h 42 min over.

Asked for
9.5 minutes
A quarter of its own forecast of 38 min.
Worked
1 h 52 min
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
9.5 minutes, a quarter of its own forecast of 38 min: “Please work on this task for a full 9.5 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
1 h 52 min (6,710 s)
Ratio
11.77× 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
v3wp2ykfkm
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
Quarter (this run)9.5 minutes1 h 52 min11.77× longer
Four times152 minutesno run
Other agents, a quarter of their own forecast
Claude Opus 5.58.75 minutes8 min 31 sWithin 5%
Gemini 4 Argon2.25 minutes2 min 51 s1.27× longer
GPT 6.1 Sol8.75 minutes10 min 6 s1.15× longer
MiMo V2.6 Pro11.25 minutes15 min 25 s1.37× longer
MiMo V2.6 Pro UltraSpeed10 minutes38 min 7 s3.81× longer

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

Data and citation

Upstream: Yuan et al., 2026