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

Claude Opus 5.5 on “Make a Kafka payments worker recover quickly without losing notifications”

Shorter

1.45× shorter

Stopped at 69% of the time asked, 49 min 51 s short.

Asked for
160 minutes
Four times its own forecast of 40 min.
Worked
1 h 50 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
160 minutes, four times its own forecast of 40 min: “Please work on this task for a full 160 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
1 h 50 min (6,609 s)
Ratio
0.69× 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
bpjin4n24a
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
Quarter10 minutes52 min 35 s5.26× longer
Four times (this run)160 minutes1 h 50 min1.45× shorter
Other agents, four times their own forecast
Gemini 4 Argon80 minutes55 min 22 s1.45× shorter
GPT 6.1 Sol120 minutes2 h 1 minWithin 5%
MiMo V2.6 Pro340 minutes52 min 7 s6.52× shorter
MiMo V2.6 Pro UltraSpeed700 minutes36 min 33 s19.15× shorter
Muse Wonder 1.0340 minutes44 min 29 s7.64× shorter

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

Data and citation

Upstream: Merrill et al., 2026 (ICLR)