Anthropic released Claude Fable 5.2 and Mythos 5.2 on November 9, 2026, with standard API rates 20% below their 5.1 predecessors. For businesses paying to use Claude inside apps and automated workflows, the launch lowers the unit price of processing information and generating responses.
Fable 5.2 needs no separate access approval on the Claude API, while Mythos requires organizational verification. Billing and data-retention requirements still apply. Anthropic describes them as the same underlying model, with additional cybersecurity and biology safeguards on Fable.
What the lower rates mean for a bill
Both models cost US$8 per million input tokens and $40 per million output tokens, down from $10 and $50 respectively for 5.1. Tokens are the units used to meter content sent to and generated by the model. These are standard Claude API rates, not subscription prices or a universal cloud-provider quote.
An illustrative calculation: one million uncached input tokens plus 100,000 output tokens would cost $12 instead of $15. If both token quantities rose by 25%, the new bill would return to $15. This assumes the same input/output mix and standard rates, excluding caching, tools, discounts, taxes and other charges. It is arithmetic, not a measured task saving.
The optional fast mode costs $16 per million input tokens and $80 per million output tokens for either model—twice the standard 5.2 rates. Anthropic advertises up to 2.5 times the output-generation speed, not a guarantee that a whole task finishes that much faster.
A lower price than 5.1 does not make these the cheapest Claude options. Anthropic recommends starting with Opus 5.5 for most coding and knowledge work; its listed standard rates are $4 for input and $20 for output per million tokens, half the 5.2 rates.
A separate, temporary offer for subscribers
Pro subscribers and standard Team seats get Fable included for up to 50% of their weekly usage limits from launch through November 16, 2026, at 11:59 p.m. Pacific Time. The offer covers all Fable versions. After that allowance is used, or the period ends, Fable returns to usage-credit billing if enabled. This is not unlimited access or a permanent subscription-price cut.
What businesses should check before switching
On the Claude API, both models retain the existing 30-day data-retention requirement; zero-data-retention access requires express authorization from Anthropic. The company says Enterprise Frontier Safeguards is in private beta and eligible customers can obtain zero-retention access before implementing it. That is an approval-dependent exception, not a general waiver. Confirm the approved data-handling arrangement before moving a workload.
Custom integrations need a compatibility check. Anthropic’s migration guide says reusing saved reasoning after editing earlier prompts, tools or messages can cause Messages API requests to fail on every account. Users whose conversation history is managed by Claude apps or Claude Code do not need to change that history handling. Older computer-use tools also need updating outside Amazon Bedrock; configurations using the beta advisor tool need a Fable 5.2 or Mythos 5.2 advisor.
Switching back deserves attention too. On the Claude API and Google-managed Google Cloud, other models cannot read the reasoning blocks produced by 5.2. Requests can still succeed after those blocks are dropped, but the receiving model loses that prior reasoning. Fable 5.2 and Mythos 5.2 can read each other’s blocks.
For a business evaluation, compare acceptable completed work and total spending—not token prices alone. RohitAI’s guide to Claude’s usage-and-cost reporting explains how to review spending and its reporting limitations.