Benchmarks retiring not because they are beaten but because they can no longer be written.
Development
A lab says it has built the last open-answer benchmark it can
Claim
Edison Scientific launched LABBench2 as the last open-answer benchmark it believes it can make, citing the difficulty of writing questions that are genuinely challenging for models.
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to February 6, 2026 (theinnermostloop.substack.com)
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.