METR staff (2.5 full-time equivalents) spent eleven weeks with access to internal Anthropic data. Most of them now believe that as of October 2026 Anthropic is improving model capabilities on well-specified tasks at "around twice the…
In the News: November 6, 2026 (Evening)
METR says Anthropic's model progress on well-specified tasks now runs at about twice its earlier pace, though Anthropic redacted most of the evidence behind it.
Evening edition
Machine-readable
Download Markdown
Story
METR estimates Anthropic's model progress has roughly doubled in pace, mostly through AI use
Read story →
Review of the "Risks from automated R&D" section in Anthropic's Risk Report (August 2026)
METR agrees catastrophic risk from Anthropic's July 15 models was low but says it is plausible they had already crossed the early warning threshold. It says the public report's "less than a factor of 2" sits against a private figure it calculates as 1.97, within its own range of 1.3 to 2.8.