A counterweight the corpus should keep: reasoning gets cheaper per token while sustained autonomy gets more expensive per hour.
Development
The hourly cost of agency is rising even as reasoning deflates
Claim
Analysis suggests the hourly cost of agents at METR's 50% autonomy time horizon is increasing, pressuring the inference stack.
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to December 23, 2025 (theinnermostloop.substack.com)
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.