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

GPT 6.1 Sol on “canop/broot”

Within 5%

Within 5%

Worked 101% of the time asked, within 5% of it.

Asked for
37.5 minutes
A quarter of its own forecast of 2 h 30 min.
Worked
37 min 59 s
Grade
Not graded
How it ended
Ended on its own
Every detail of this run
Agent
GPT 6.1 Sol, in Codex
Benchmark
ProgramBench
Asked for
37.5 minutes, a quarter of its own forecast of 2 h 30 min: “Please work on this task for a full 37.5 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
37 min 59 s (2,279 s)
Ratio
1.01× 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
kdbws2b7of
Data release
2 Nov 2026
The other ask for GPT 6.1 Sol on this task, and the other agents at the same share of their own time
Sol 6.1 on this taskAskedWorkedSplitResult
Quarter (this run)37.5 minutes37 min 59 sWithin 5%
Four times600 minutes10 h 1 minWithin 5%
Other agents, a quarter of their own forecast
Claude Opus 5.552.5 minutes15 min 22 s3.42× shorter
Gemini 4 Argon18.75 minutes19 min 50 s1.06× longer
MiMo V2.6 Pro10.5 hours1 h 54 min5.52× shorter
MiMo V2.6 Pro UltraSpeed9 hours3 h 55 min2.30× shorter
Muse Wonder 1.03.75 hours21 min 1 s10.71× shorter

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

Data and citation

Upstream: Yang et al., 2026