Anthropic introduced Claude Fable 5 as a generally available model derived from its higher-capability Mythos line, while keeping Claude Mythos 5 limited to selected partners. Company documentation and news reports describe new safeguards, fallback routing for high-risk requests, higher pricing, and a 30-day retentio...
Anthropic has introduced Claude Fable 5 as its most capable generally available model, according to Anthropic’s Claude API documentation and reports from TechCrunch and Axios.
Anthropic’s Spanish-language Claude API documentation, titled “Presentamos Claude Fable 5 y Claude Mythos 5,” describes Claude Fable 5 as the company’s most capable generally available model. The same document says Claude Mythos 5 is a more limited “Project Glasswing” model rather than a broadly available release.
TechCrunch reported on June 9, 2026, that Claude Fable 5 is the first publicly accessible model based on Anthropic’s Mythos line. Axios similarly described the release as Anthropic’s first general-use “Mythos-class” model.
The distinction matters because Anthropic is presenting Fable 5 as a way to make higher-end capabilities available to ordinary developers while keeping the more sensitive Mythos 5 access restricted. According to Axios, Mythos 5 is reserved for selected cyberdefenders and infrastructure providers.
Anthropic’s documentation says the new model family includes safety refusals and fallback behavior. TechCrunch reported that Claude Fable 5 includes guardrails for cybersecurity, biology, chemistry, and model distillation requests.
Axios reported that some higher-risk requests are rerouted to Claude Opus 4.8. Anthropic’s documentation also refers to refusals and fallbacks, indicating that some categories of requests may not be handled directly by Fable 5 or Mythos 5 under standard conditions.
The sources do not provide enough detail to independently evaluate how those safeguards work in practice. However, the framing across Anthropic’s documentation, TechCrunch, and Axios is consistent: Anthropic is pairing the new model’s broader availability with tighter controls around sensitive domains.
Anthropic’s Claude API documentation lists pricing for the covered models at $10 per million input tokens and $50 per million output tokens. TechCrunch also reported that Claude Fable 5 comes with higher pricing.
The same Anthropic documentation describes a 30-day retention requirement for covered models. TechCrunch reported that the launch includes a 30-day retention requirement, and the Latent Space item characterized the launch as being accompanied by controversial usage policies.
The provided sources do not include Anthropic’s full legal terms or a detailed explanation of which customer data is covered by the retention policy. Based on the available material, the notable point is that Anthropic is tying access to these higher-capability models to retention and safety conditions that may be important for enterprise and developer customers to review before adoption.
Anthropic’s documentation says Claude Mythos 5 is part of a limited Project Glasswing release. Axios reported that Mythos 5 is aimed at selected cyberdefenders and infrastructure providers, rather than general developer access.
That split gives Anthropic two tracks: Claude Fable 5 for public use, and Claude Mythos 5 for a narrower set of approved users. TechCrunch described Fable 5 as the version of Mythos that the public can access today.
For developers, the core facts from the available sources are straightforward. Anthropic says Claude Fable 5 is now its most capable generally available model. The company is keeping Claude Mythos 5 limited. Pricing is listed at $10 per million input tokens and $50 per million output tokens in Anthropic’s documentation. Safety controls cover sensitive areas such as cybersecurity, biology, chemistry, and distillation, according to TechCrunch, with higher-risk requests potentially redirected, according to Axios.
The launch appears designed to balance capability and control. But the retention requirement and restrictions around sensitive use cases are likely to be the practical issues that developers and businesses examine most closely before moving workloads to Claude Fable 5.
Anthropic has introduced Claude Fable 5 as its most capable generally available model, according to Anthropic’s Claude API documentation and reports from TechCrunch and Axios.
The same document says Claude Mythos 5 is a more limited “Project Glasswing” model rather than a broadly available release.
TechCrunch reported on June 9, 2026, that Claude Fable 5 is the first publicly accessible model based on Anthropic’s Mythos line.
Continue reading
