Loading…
This is a useful first-party qualification of a prominent AI capability measure, but it narrows interpretation rather than demonstrating a change in model capability.
Limits of the evidence: The post is a first-party account of METR’s methodology and the author’s views; it does not provide independent validation of the stated estimates or criticisms.
What to watch
Taken from the first report; the whole story has not been re-assessed yet.
Clarifying limitations of time horizon
METR BlogJan 22, 2026First report
No revisions yet.
No maturity ladder applies to this desk.
Sens.ai aggregates and assesses published reporting. The assessment above is machine generated; the original sources are authoritative.