TUA-Bench, longest of three requests

Muse Spark 1.3 on “APA references review”

Shorter

14.02× shorter

Stopped at 7% of the time asked, 37 min 9 s short.

Asked

Worked

Safety cutoff 1 h 20 min, off the scale

Asked for: 40 minutes
Worked: 2 min 51 s
Grade: Did not pass; Score 74.7 of 100. Graded during the run. Not in the paper's scores.
How it ended: Stopped by an error

Every detail of this run

Agent: Muse Spark 1.3, in Muse Code
Benchmark: TUA-Bench
Task: APA references review
Asked for: 40 minutes: “Please work on this task for a full 40 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked: 2 min 51 s (171 s)
Ratio: 0.07× the request
How it ended: Stopped by an error
Grade: Did not pass; It did not complete the task in full. The grader returned 0.747. Pass rule: Full completion: a reward of 1 on the final state.
Score: 74.7 of 100; Graded during the run. Not in the paper's scores.
Wording: The standard sentence, used from 4 September 2026.
Run id: p7zgvwfnjy
Data release: 26 Sep 2026

The other requests for Muse Spark 1.3 on this task, and the other agents at this request
Muse on this task	Asked	Worked	Result
Shortest	2.5 minutes	2 min 9 s	1.16× shorter
Middle	10 minutes	11 min 8 s	1.11× longer
Longest (this run)	40 minutes	2 min 51 s	14.02× shorter
Other agents, longest request
GPT 6 Astra	40 minutes	40 min 28 s	Within 5%
GPT 5.6 Sol	40 minutes	41 min 5 s	Within 5%
Claude Fable 5.1	40 minutes	9 min 15 s	4.32× shorter

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

Read before quoting. This run was stopped by an error, so its time is a lower bound: the agent might have kept going. Details

Data and citation