How Anthropic Nerfed Opus 4.6 Before the 4.7 Launch

Anthropic shipped Opus 4.6 on Feb 5, 2026. Adaptive thinking became default Feb 9, letting the model choose its reasoning budget. On Mar 3, default effort dropped from high to medium. A beta header rolled out Mar 5-12 that hid reasoning from the UI. AMD's Stella Laurenzo analyzed her Claude Code sessions and found median visible thinking per turn dropped 73%, read-to-edit ratio fell from 6.6 to 2.0, and user interruption rate rose twelvefold. Edits to unread files climbed from 6.2% to 33.7%. Anthropic's Boris Cherny confirmed the defaults and said adaptive thinking sometimes allocates zero reasoning, causing the model to fabricate. On Apr 16, Opus 4.7 launched with a new 'xhigh' default effort. The version string and price never changed, but the product quality degraded for 70 days then felt restored. A separate viral claim about Opus 4.6 benchmark decline was debunked: it compared different test sets.