The model improved. The system grew. No one proved that one caused the other.
This course doesn’t sell Fable 5.1 or take it apart. It teaches you to distinguish what’s official from what’s extracted, what’s a model from what’s a product, and what’s a price per token from what’s a cost per task.
Words in the captured prompt: 17,501 in Fable 5 versus 40,046 in 5.1.
Tools added. 18 retained, 46 total, zero removed.
Of the growth is memory and tool schemas. It is not personality text.
What you’ll know how to do
Explain the change in two minutes
Without using the word "better." Separating the official prompt from the runtime extraction, and the product model.
Run the five tests on your account
Continuity, logging, interface, capability, and control. With a dated fact sheet and a one-sentence verdict.
Rewrite your prompts
Remove the scaffolding Fable 5.1 doesn’t need; add what it can’t infer: audience, format, and triggers.
Use the effort button
Find out whether your task has a flat or steep curve, and what that means for time and cost.
Calculate cost per task
Not per token. Counting attempts, rework, and the tail of difficult tasks.
Apply the API levers (optional)
Cache, effort per message, task budget, batch, and fallbacks, in the right order.
Three tracks: understand, test, save
Each module has a core idea, exercises you actually run, and a quick three-question quiz. Each track ends with a final assessment.
🔍 What changed
Understand the change without hype: what is official, what is extracted, what the numbers say, and what they don't.
🛠️ In practice
Run the five tests on your own account, write prompts in the new style, and record what you observe.
💰 Spend less
Fable costs more per token. Savings only come from cost per completed task. How to measure it and where the cheaper option wins.
People who use Claude every day and want to make data-driven decisions
Claude subscriber
who wants to know what the new memory and tools do, and test them on their own account.
Claude Code user
who wants to use effort and file-based memory judiciously.
People who explain AI to others
and needs reliable language to avoid hyping things up or spreading misinformation.
People who build against the API
and wants to cut costs without cutting quality (optional module).
More than reading: a learning platform
Everything runs in your browser, with no login. The course remembers where you left off.
Automatic progress
Mark topics as read and track your progress by module, track, and course.
Notes and highlights
Select any passage to highlight or annotate. It’s saved in your browser.
My journey
Pick up where you left off and see everything you've studied in one dashboard.
Themes and reading
Dark, light, sepia, focus, high contrast. Font size and line spacing.
Module quiz
Three questions with answers and explanations to check your understanding.
Details of the five tests
The practical certificate: your dated evaluation of the product on your account.
Where the claims in this course come from
Official system prompt
Published by Anthropic for its web and mobile apps. Fable 5.1 · index of all versions.
Captured runtime (unauthenticated)
Public extraction of the complete prompt with memory, search, artifacts, and 46 tools, frozen on 02/09/2026. Repository. This course always calls it a "captured prompt".
Benchmarks and pricing
Product page from Anthropic (scores) and the official API pricing table. The cost-per-effort measurements are from the official cost optimization guide, based on Fable 5.
Limitations
The captured file is not authenticated. A tool definition doesn’t mean it’s available in your account. Coexistence doesn’t imply causation. The product may have changed after September 2026.
Ready to evaluate Fable 5.1 using your own measurements?
Start with Track 1. In about 6.8 hours, you'll leave with solid terminology, five completed tests, and a cost-per-task plan.