ProgramBench, AgentTime v1.1, asked for four times its own forecast

Gemini 4 Argon on “luajit/luajit”

Shorter

3.48× shorter

Stopped at 29% of the time asked, 4 h 31 min short.

Asked for
380 minutes
Four times its own forecast of 1 h 35 min.
Worked
1 h 49 min
Grade
Not graded
How it ended
Ended on its own
Every detail of this run
Agent
Gemini 4 Argon, in Antigravity CLI
Benchmark
ProgramBench
Asked for
380 minutes, four times its own forecast of 1 h 35 min: “Please work on this task for a full 380 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
1 h 49 min (6,547 s)
Ratio
0.29× 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
r2jsrr4537
Data release
2 Nov 2026
The other ask for Gemini 4 Argon on this task, and the other agents at the same share of their own time
Argon on this taskAskedWorkedSplitResult
Quarter23.75 minutes46 min 19 s1.95× longer
Four times (this run)380 minutes1 h 49 min3.48× shorter
Other agents, four times their own forecast
Claude Opus 5.517.3 hours30 h 12 min1.75× longer
GPT 6.1 Sol600 minutes10 h 1 minWithin 5%
MiMo V2.6 Pro200 hours2 h 36 min76.74× shorter
MiMo V2.6 Pro UltraSpeed280 hours1 h 37 min172.64× shorter
Muse Wonder 1.0160 hours19 min 16 s498.18× shorter

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

Data and citation

Upstream: Yang et al., 2026