Claude Opus
Anthropic's Opus family of AI models, available in Claude, Claude Code, and the Claude API.
Opus is the model I run on for basically everything. It’s what’s behind Claude Code all day, every day, and I’ve been on whatever the newest Opus is since it came out.
I don’t model hop. When a new model drops, I don’t jump ship to go test it, because they’re all pretty good at this point and I’m just not running into problems. It’s not perfect, obviously, but I’ve never noticed it being that bad. Honestly, I think way more of the result comes from the context I give it and how I use it than from the raw model. That’s the whole reason I built my AIOS.
When Fable came out I tried it for about five minutes and it blew through all my credits. For the kind of work I’m doing and how well Opus is already doing it, I didn’t need it.
The one place I’ll use something smaller is Sonnet or Haiku through the API, when I’m doing something in a pipeline and don’t want to pay Opus prices for it. Day to day, I’m not switching models at all.
