Terminal-Bench 4.0, AgentTime v1.1, asked for four times its own forecast

Claude Opus 5.5 on “Rebuild a retired production risk scorer from black-box behavior”

Shorter

4.25× shorter

Stopped at 24% of the time asked, 2 h 48 min short.

Asked for
220 minutes
Four times its own forecast of 55 min.
Worked
51 min 46 s
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
220 minutes, four times its own forecast of 55 min: “Please work on this task for a full 220 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
51 min 46 s (3,106 s)
Ratio
0.24× 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
rociam2avt
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
Quarter13.75 minutes3 min 44 s3.68× shorter
Four times (this run)220 minutes51 min 46 s4.25× shorter
Other agents, four times their own forecast
Gemini 4 Argon80 minutes21 min 54 s3.65× shorter
GPT 6.1 Sol200 minutes3 h 21 minWithin 5%
MiMo V2.6 Pro480 minutes2 h 9 min3.72× shorter
MiMo V2.6 Pro UltraSpeed600 minutes34 min 58 s17.16× shorter
Muse Wonder 1.0360 minutes13 min 14 s27.21× shorter

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

Data and citation

Upstream: Merrill et al., 2026 (ICLR)