Your Claude guides have a sell-by date

Your Claude Guides Have A Sell-By Date

Your Claude guides have a sell-by date

Read time: 3 minutes

Every Claude guide you saved this spring is now wrong somewhere. It names a model that no longer leads, or a thinking setting current models refuse.

Nobody stamped a date on it.
You followed it anyway.

You probably tried the obvious fixes. Newer guide, newer creator. Those age at the same speed, because the lineup turned over twice between March and September.

The guide isn't the problem. The missing date is. A page that can't tell you when it expired keeps sounding confident long after it stopped being true.

So here is the current state, dated 4 September 2026, and a check you can run on any guide in under a minute.

In this issue

→ 1. The lineup, right now
→ 2. Opus 5 is the daily default
→ 3. Extended thinking is gone
→ 4. Effort is the control now
→ Do this now: the three-check test

1. The lineup, right now

Four models. Everything else is legacy. Prices and cutoffs are in the table above.

Fable 5.1 shipped on 1 September. Fable 5, Opus 4.8, 4.7, 4.6, and Sonnet 4.6 still run. None of them is the answer any more.

Trap: any guide calling Opus 4.x the newest model is from spring. Any guide calling Fable 5 the top is from August.

2. Opus 5 is the daily default, and Anthropic says so

Anthropic tells you to start in the middle, not at the top. Same price as Opus 4.8, and the freshest knowledge of the mid-range, May 2026.

Anthropic's own advice, from the models page: start with Opus 5 for most workloads. Reach for Fable 5.1 only for demanding reasoning and long agent runs.

3. Extended thinking is gone from current models

Extended thinking with a token budget isn't accepted on current models. On Fable 5.1, Opus 5, and Sonnet 5, thinking is already on.

Claude decides when and how deeply to think, and the text stays hidden unless you ask for it.

On the older 4.x models it's off until you switch on adaptive thinking. The manual budget slider your guide shows you belongs to a generation that is now legacy.

Trap: "turn on extended thinking" was correct in May. On a current model it's either a no-op or an error.

4. Effort is the control that replaced it

Effort is the cost dial now. Five levels: max, xhigh, high, medium, low. High is the default and behaves the same as setting nothing.

Three things about it your guide doesn't say:

  • It scales every token, not only thinking. Tool calls, arguments, explanations. Low effort means fewer calls and no preamble.
  • On Opus 5 it doesn't shorten the answer. Anthropic says so directly. For less text, ask for less text.
  • It moved the levers. Shorter chats and model downgrades still help. They are the second fix now.
    Effort is the first.

One more number that moved underneath you. The tokenizer changed at Opus 4.7, and every model since uses it. Same text, roughly a third more tokens.

☐ If your bill rose without your usage changing, start there.

Do this now: the three-check test

Run this on the next Claude guide you open. Under a minute.

  1. Find the date. No date is itself an answer.
  2. Check the model names against the four above. Opus 4.x as newest means spring. Fable 5 as top means August.
  3. Search the page for the word effort. If it talks about cost or thinking and never says effort, it predates the control that governs both.

Three checks, and confident advice about a product that moved stops costing you. You keep the guides that held. You stop following the ones that quietly expired.

This issue is the working version. The full build is in this week's Substack edition. That means the complete model table, Anthropic's effort guidance quoted, the tool-use change that breaks Fable 5 migrations, and the six-step check.

If you want the whole thing:

Read the expanded edition →


PS - Whenever you’re ready, here are 3 ways I can help you:

  1. Free Resources - Get swipe files, templates & AI prompts to accelerate your growth.
  2. Online Courses – Learn step-by-step from proven frameworks to build and scale your creator business.
  3. Book 1:1 Strategy Call - Being stuck is a choice. In 30 minutes, we’ll map your next 3 moves to get you unstuck. No cost, no strings. Creators who win don’t wait.

Have questions? Hit reply to this email, and we'll help out!

AI for Leverage

You are already good at the work. The output does not show it yet. Every week I hand you one AI system that gives hours back, already running in my own business, with the numbers attached.