Claude Fable 5.1 Is Here: What Has Changed?
Anthropic unveiled Claude Fable 5.1 on September 1. It strengthens coding and long-running task execution and cuts cache-read pricing by 75% compared with the previous model. Official product information
From Writing Code to Validating It
Anthropic presents implementing features across a codebase, code review, and performance improvements as use cases for Fable 5.1. It also highlights writing tests itself and inspecting screens to check whether the implementation meets its goals. Extended tasks involving multiple tools are another major use case. Feature overview
The Pricing Change Is in Cache Reads
API prices per million tokens:
| Item | Price |
|---|---|
| Input | $10 |
| Output | $50 |
| Cache reads | $0.25 |
<p class="blog-price-source"><a href="https://www.anthropic.com/claude/fable" data-i18n="blog.price.source">Official pricing</a><span data-i18n="blog.price.checked"> · Checked September 5, 2026</span></p>
Cache-read pricing is 75% lower than Fable 5. This does not mean the total bill always falls by 75%. Actual savings depend on how much of the workload involves repeatedly reading cached content. Pricing information
Where Can You Use It?
It is available to Claude Pro, Max, Team, and Enterprise users. Developers can also access it through Claude Platform, AWS, Google Cloud, and Microsoft Foundry. Its Claude API model identifier is claude-fable-5-1. It is worth checking subscription access and API billing separately. Availability information
Applying It to Development Work
One useful way to compare a new model is to give it a familiar bug. Provide the same reproduction steps and code, then compare its explanation of the cause, the scope of the fix, and the test results. A practical criterion is whether it reduces the total working time, including review, rather than simply whether it answers quickly.
This is a usage suggestion from GamZip, not an experimental result guaranteeing Fable 5.1's performance.