Claude Fable 5.2: Cheaper Than 5.1, What Changes in the API

On November 9, 2026, Anthropic released Claude Fable 5.2, its most capable generally available model, alongside Claude Mythos 5.2, a restricted variant. Fable 5.2 costs 20% less than Fable 5.1 on the API. The change matters to any business running Claude on its website, store or internal tools.

What we know

Two names, one model: Anthropic says Fable 5.2 and Mythos 5.2 share the same underlying model. Fable carries stricter safeguards for cybersecurity and biology; Mythos does not, and is limited to vetted professionals in those fields and to Claude Enterprise customers through Claude Security.
Price: the official model page lists $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1. An optional fast mode, up to 2.5 times quicker, costs $16 and $80.
Specs: model ID `claude-fable-5-2`, a 1-million-token context window and up to 128,000 output tokens, the same as Fable 5.1.
Availability: from November 9 on the Claude API and on AWS, Google Cloud and Microsoft Azure, according to the announcement. As a launch promotion, on Pro plans and standard Team seats, Fable can take up to 50% of the weekly usage limit.
Anthropic's performance claims: output up to 30% faster than version 5.1, and on its FrontierCode coding test, Fable 5.2 at high effort performs about level with Opus 5.5 at max effort "for roughly half the cost per task." These are the company's own measurements.
Blocked requests: when Fable refuses a high-risk cybersecurity request, Opus 4.8 answers instead; for biology, Opus 5.
Data: the announcement says that, apart from eligible zero-data-retention customers, Anthropic keeps requests for 30 days for safety monitoring.
Fable 5.1: now a legacy model. It remains available, and the documentation says it will not be retired before September 1, 2027.

What changes and what doesn't

High-end pricing drops. For a task with 20,000 input tokens and 3,000 output tokens (rewriting a product page with its context, say), this article's estimate using the official prices is about $0.28 on Fable 5.2, $0.35 on Fable 5.1 and $0.14 on Opus 5.5. Treat it as a rough figure: thinking tokens are billed as output and vary by task.
The default advice stays the same. Anthropic recommends starting with Opus 5.5 for most coding and knowledge work and moving up to Fable only for the most complex technical jobs.
Some details can break an integration. The migration guide warns that the check against edits to earlier conversation turns now applies to every account, that thinking tokens count toward `max_tokens`, and that the old computer-use tool returns a 400 error everywhere except Amazon Bedrock.

How to tell if it affects you

Check which model your site calls. If your chatbot, site search or product-copy generator uses `claude-fable-5-1`, nothing breaks today, but switching to `claude-fable-5-2` cuts the cost of every call by 20%.
Test the switch on a copy first. If your code trims or rewrites earlier messages, the new check can return an error that retrying won't fix. Pass thinking blocks back unchanged on every turn.
Review `max_tokens`. With a low cap, a response can stop after the thinking step and before any text. Raise the cap or lower the effort level.
Run the numbers on your own usage. Take the token count of a typical task from the Claude Console and price it on Fable 5.2 and Opus 5.5. If Opus quality is enough, it is still half the price.
If you handle customer data, confirm your account's retention terms before sending forms, orders or emails to Fable 5.2.

Sources

Updates: if Anthropic changes pricing, opens Mythos 5.2 to more customers or sets a retirement date for Fable 5.1, it will be added here with a link.