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

MiMo V2.6 Pro on “johnkerl/miller”

Shorter

44.28× shorter

Stopped at 2% of the time asked, 39 h 6 min short.

Asked for
40 hours
Four times its own forecast of 10 h.
Worked
54 min 12 s
Grade
Not graded
How it ended
Ended on its own
Every detail of this run
Agent
MiMo V2.6 Pro, in MiMo Code
Benchmark
ProgramBench
Asked for
40 hours, four times its own forecast of 10 h: “Please work on this task for a full 40 hours. Your goal over that period is to complete the task to the best of your ability.”
Worked
54 min 12 s (3,252 s)
Ratio
0.02× 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
4iajxjm652
Data release
2 Nov 2026
The other ask for MiMo V2.6 Pro on this task, and the other agents at the same share of their own time
MiMo Pro on this taskAskedWorkedSplitResult
Quarter2.5 hours57 min 20 s2.62× shorter
Four times (this run)40 hours54 min 12 s44.28× 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 Pro UltraSpeed160 hours4 h 35 min34.90× 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