December Awakening
Everything accelerated, again, in Nov-Dec-Jan. Lots of people are noticing.
We passed some kind of crazy milestone recently, one that is only becoming obvious in retrospect. I felt it, and wondered if others had sensed it too.
It turns out they had.
What I’ve Been Saying
In Leveling Up Claude Code with Boris Cherny’s Help:
Claude Opus 4.5 arrived in November, powering up Claude across all its surfaces—Claude.ai, Claude Desktop, and Claude Code. So smart, such encyclopedic knowledge.
In The Workflow:
Moving from “I wish I could do something about X” to “I can definitely now do something about X—and Y and Z and A-B-C,” is a huge deal. AI is the enabler, but more specifically, the November arrival of Claude Opus 4.5 was, for me, the tipping point. That model crossed some kind of “smart enough” line where there was literally no workflow challenge it could not take on. (Things are now even better with Opus 4.6.)
What Others Are Saying
I’ll put Doug O’Laughlin from SemiAnalysis up front because he’s the one who calls this phenomenon “The Awakening” in this Latent Space podcast with Shawn “Swyx” Wang:
O’Laughlin is also the guy who detected and charted the insane trend whereby Claude Code co-authored commits now represent a substantial and exponentially-growing percentage of all commits on GitHub:
Swyx himself covers the phenomenon thoroughly in this Latent Space AI news post:
Swyx even created a dedicated site for it, wtfhappened2025.com:
Ethan Mollick wrote on LinkedIn in December about Opus 4.5’s off-the-chart performance on METR benchmarks:
Here’s METR’s updated chart for Opus 4.6, yet another exponential outlier:
David Heinemeier Hansson (DHH) responding to Karpathy:
Boris Cherny on Twitter in December:
Pretty convincing evidence, eh?
My Take on What’s Powering This
- The arrival of Claude Opus 4.5 in November began the exponential
- Some kind of significant Claude Code agent harness improvements in December ratcheted up the curve
- Opus 4.6 in January, even more so
