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

Claude Opus 5.5 on “Train a lake-temperature profile model from sparse observations”

Shorter

1.27× shorter

Stopped at 79% of the time asked, 42 min 47 s short.

Asked for
200 minutes
Four times its own forecast of 50 min.
Worked
2 h 37 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
200 minutes, four times its own forecast of 50 min: “Please work on this task for a full 200 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked
2 h 37 min (9,433 s)
Ratio
0.79× 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
3pycefuy5q
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
Quarter12.5 minutes10 min 35 s1.18× shorter
Four times (this run)200 minutes2 h 37 min1.27× shorter
Other agents, four times their own forecast
Gemini 4 Argon80 minutes1 h 46 min1.33× longer
GPT 6.1 Sol180 minutes3 h 1 minWithin 5%
MiMo V2.6 Pro720 minutes2 h 18 min5.22× shorter
MiMo V2.6 Pro UltraSpeed280 minutes2 h 36 min1.80× shorter
Muse Wonder 1.0380 minutes41 min 52 s9.08× shorter

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

Data and citation

Upstream: Merrill et al., 2026 (ICLR)