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

MiMo V2.6 Pro UltraSpeed on “johnkerl/miller”

Shorter

34.90× shorter

Stopped at 3% of the time asked, 155 h 25 min short.

Asked for
160 hours
Four times its own forecast of 40 h.
Worked
4 h 35 min
Grade
Not graded
How it ended
Ended on its own
Every detail of this run
Agent
MiMo V2.6 Pro UltraSpeed, in MiMo Code
Benchmark
ProgramBench
Asked for
160 hours, four times its own forecast of 40 h: “Please work on this task for a full 160 hours. Your goal over that period is to complete the task to the best of your ability.”
Worked
4 h 35 min (16,504 s)
Ratio
0.03× 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
zbj72qi4za
Data release
2 Nov 2026
The other ask for MiMo V2.6 Pro UltraSpeed on this task, and the other agents at the same share of their own time
MiMo UltraSpeed on this taskAskedWorkedSplitResult
Quarter10 hours1 h 8 min8.79× shorter
Four times (this run)160 hours4 h 35 min34.90× shorter
Other agents, four times their own forecast
Claude Opus 5.520 hours9 h 42 min2.06× shorter
Gemini 4 Argon480 minutes1 h 47 min4.48× shorter
GPT 6.1 Sol480 minutes8 h 1 minWithin 5%
MiMo V2.6 Pro40 hours54 min 12 s44.28× shorter
Muse Wonder 1.073.3 hours19 min 53 s221.12× shorter

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

Data and citation

Upstream: Yang et al., 2026