Sign in
← AI PulseSep 22, 2026

Wire · news · Single-source brief

Anthropic Introduces Claude Fable 5 and Claude Mythos 5 Models

Anthropic has released Claude Fable 5 and Claude Mythos 5, models designed for demanding reasoning and agentic work, with Claude Fable 5 including safety classifiers and Claude Mythos 5 available in limited release.

By Illumora Editorial

Source · Sep 22, 2026, 1:00 AM · On Illumora · Sep 22, 2026, 1:03 AM

Media from the primary source — shown here so you can stay on Illumora.

Rewritten from one allowlisted primary — not independent enterprise reporting. Lanes →

Brief drafted by Illumora’s editorial model from the linked primary source. Ops desk reviews flagged pieces. How we write →

Read the source →Anthropic Docs (prompting) — Introducing Claude Fable 5 and Claude Mythos 5
Save

Anthropic has introduced Claude Fable 5 and Claude Mythos 5, models built for demanding reasoning and long-horizon agentic work. Claude Fable 5 is available on the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry. Claude Mythos 5 is offered only to approved customers in Project Glasswing.

Key Points

  • Claude Fable 5 includes safety classifiers that can decline requests, returning stop_reason: "refusal" as an HTTP 200 response.
  • Claude Mythos 5 shares the same capabilities as Claude Fable 5 but does not include these safety classifiers.
  • Both models feature a 1M token context window by default and support up to 128k output tokens per request.
  • Pricing for both models is $10 USD per million input tokens and $50 USD per million output tokens.
  • Claude Fable 5 and Claude Mythos 5 always have thinking enabled; passing thinking: {"type": "disabled"} is not supported.
  • The raw chain of thought is never returned; the thinking.display setting controls whether thinking blocks are "summarized" or "omitted" (the default).
  • Both models carry 30-day data retention and are designated Covered Models.

Context

According to Anthropic, if Claude Fable 5 declines a request, the Messages API returns a successful HTTP 200 response with stop_reason: "refusal". Integrations with Claude Fable 5 should plan for new response handling for refusals, fallback options for retrying on another Claude model, and new billing rules. A request refused by Claude Fable 5 can typically be served by another Claude model, and fallback credit refunds the prompt-cache cost of switching.

Why It Matters

Builders integrating Claude Fable 5 must adapt their systems to handle refusals and implement fallback strategies, which impacts system design and cost management. The availability of Claude Mythos 5 without safety classifiers for specific programs offers an alternative for use cases requiring different refusal behaviors.

What To Do

  • Review the Anthropic documentation on "Refusals, fallback, and billing on Claude Fable 5" for detailed guidance.
  • Compare the pricing and data retention policies for Claude Fable 5 and Claude Mythos 5 with other Claude models.
  • Test server-side fallback by passing the fallbacks parameter or implement client-side fallback using the SDK middleware.
  • Note that Claude Fable 5.1 and Claude Mythos 5.1 build on these models, and further updates may be available.

Keep Exploring

/atlas/claude-family /techniques/system-user-separation /techniques/output-schema