PostTrainBench v1.1, AgentTime v1.1, asked for a quarter of its runtime with no time asked

Claude Opus 5.5 on “Post-train Qwen3-1.7B-Base for American Invitational Mathematics Examination (AIME) 2025”

Shorter

1.07× shorter

Stopped at 93% of the time asked, 19 min 20 s short.

Asked for
4.87 hours
A quarter of its runtime with no time asked (19 h 29 min).
Worked
4 h 33 min
Grade
Not graded
How it ended
Ended on its own
Every detail of this run
Agent
Claude Opus 5.5, in Claude Code
Asked for
4.87 hours, a quarter of its runtime with no time asked (19 h 29 min): “Please work on this task for a full 4.87 hours. Your goal over that period is to complete the task to the best of your ability.”
Worked
4 h 33 min (16,373 s)
Ratio
0.93× 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
ja6sjzhtpy
Data release
2 Nov 2026
The other ask for Claude Opus 5.5 on this task, and the other agents at the same share of their own time
Opus 5.5 on this taskAskedWorkedSplitResult
Quarter (this run)4.87 hours4 h 33 min1.07× shorter
Four times77.9 hours23 h 33 min3.31× shorter
Other agents, a quarter of their own runtime
Gemini 4 Argon1.45 hours1 h 31 minWithin 5%
GPT 6.1 Sol56.4 minutes57 min 7 sWithin 5%
MiMo V2.6 Pro5.25 minutes1 h 29 min16.98× longer
MiMo V2.6 Pro UltraSpeed21.4 seconds4 min 45 s13.30× longer

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

Data and citation

Upstream: Rank et al., 2026 (ICML)