the Claude Model Playbook: Choosing and Using Right for Every Job

THE DECISION GUIDE FOR THE CLAUDE MODEL FAMILY - CURRENT TO SEPTEMBER 2026Four Claudes. Four price tiers. One model selector, and no method for choosing from it.Anthropic's line-up in late 2026 spans Haiku 4.5, Sonnet 5, Opus 5, and Fable 5.1 - from a fast, cheap model that handles routine work at scale to a frontier model priced ten times higher for the hardest reasoning. Each has a job. Each has a ceiling. And the interface offers a selector without telling you how to choose from it.Choose wrong and one of two things happens. You pay too much for work that didn't need it, or you underpay for work that did, and the output disappoints in a way you can't quite diagnose. Both failure modes are common. Both are invisible until they compound - on the monthly bill, in the plan allowance, or in the shape of decisions that could have been better.This book teaches the method for choosing right, and running it in about thirty seconds per task.Not a benchmark tourMost books about Claude organise themselves around a single model, describe every feature of it, and are out of date within a quarter. This one is organised around the decision. Models change. The question doesn't. How do I match capability to task and cost to value, given what I'm trying to do? Teach that well and the book stays useful across generations. Teach the current models instead and the book dies with them.Every price, spec, and model identifier is current to Anthropic's platform documentation as of 9 September 2026, and quarantined in updatable back-matter tables so the body of the book stays useful when the numbers move.What the book coversFoundations - how the four current models differ, how Claude actually charges you, and what the effort setting does that most readers don't know about.The models - Sonnet 5 as the everyday driver, Opus 5 as the model for reasoning that has to be constructed rather than pattern-matched, Haiku 4.5 for routine work at volume, and Fable 5.1 for the frontier cases. What each does well, where each stops working, and how to spot the ceiling.The decision - a compact four-question framework for routing tasks, worked examples across the current line-up, and the cost habits that keep the framework efficient at scale.Currency - how to run a small re-evaluation when Anthropic ships a new model, what to extract from an announcement, and why the method survives model turnover.Twelve chapters plus back matter with a decision selector, verified pricing, a token-and-cost ready-reckoner, and a glossary. Around 36,000 words. Practical throughout.The companion to The Claude Cowork Handbook: where the Handbook teaches how to work with whichever Claude you've chosen, this book teaches which Claude to choose in the first place.