Picking the right Claude model takes about ten seconds once you know the rule, and on September 1 Anthropic made it easier by saying its own answer out loud.
They shipped Fable 5.1 that day and moved their default recommendation to the middle model, Opus 5.
That's the news if you run a business. Everything else about Fable 5.1 is for developers.
Save money by skipping the newest model
Fable 5.1 sits at the top of the range and it's API and Claude Code only for now. You won't see it in the app picker.
What moved is for the people building on it: reading a long context again now costs 75% less. The headline price didn't budge, still $10 in and $50 out per million tokens.
So it's a heavy-agent model. If you're running a business rather than a lab, you don't need it, and I'd say that plainly.
Introducing The First Agentic CRM
Get revenue agents, workflows, and automations across every stage of your motion. Access customer data in real time through Attio's web app, MCP, API, and SDK.
Then Ask Attio anything about your business and get instant answers.
It's the CRM that runs the work behind every win.
Pick the right model, job by job
Sonnet 5 for the daily stuff. Emails, summaries, first drafts, quick edits. It's been the default since June 30 and it's fast.
Opus 5 when the job has more than three steps. A sales page from a brief, a contract read, a research pull across twenty sources, anything in Cowork that touches your files. This is what Anthropic now recommends as the default for real work, and since July 28 it's half the price it used to be.
Fable only if somebody you pay is building agents on the API. If that's not you, forget it exists.
The mistake I see most is people running Opus for a two-line email and Sonnet for a fifty-page analysis, then blaming the model.
Set it once and never touch the picker
In the app, the model picker sits at the top of the chat (not buried in settings, which is where everyone looks first). In Cowork you set it per session. And in Projects you can pin one, which is the move: a Daily project on Sonnet, a Deep work project on Opus, and you never think about it again.
If you want Claude to tell you which one a job needs, ask it:
Before you start, tell me in one line whether this task needs Sonnet or Opus, and why. Then do it.
Task: [paste the job]I've been surprised how often it says Sonnet.
Picking the model is a two-minute decision. The setup around it is the real work: which project holds what, which tools are connected, what runs on a schedule. That's what turns Claude from a chat window into the thing running your business, and it's what the community builds, one setup a month, files included.
Your business, run on Claude. One owner, a full team's output.
Quick one.
Which model do you actually run day to day? Hit reply, I'm curious whether people switched after July.
Talk soon,
Zephyr
