Anthropic releases Claude Fable 5, its first publicly available Mythos-class AI model

Jun 9, 2026

Anthropic has released Claude Fable 5, the first publicly available version of its powerful Mythos-class AI model. Previously withheld due to cybersecurity risks, the model's release is enabled by new safeguards that block high-risk queries and fall back to an older model. Priced at double its predecessor, Fable 5 demonstrates state-of-the-art performance in software engineering and complex analysis, drawing praise from early customers.

Claude Fable 5 launch

▪Anthropic launched Claude Fable 5, its first publicly available Mythos-class AI model, on June 9, 2026.
▪Pricing for Claude Fable 5 is $10 per million input tokens and $50 per million output tokens.
▪Alongside Fable 5, Anthropic also launched Claude Mythos 5, an updated version of its advanced model for trusted partners.
▪This price is double the rate for the previous model, Claude Opus 4.8, and less than half the price of Claude Mythos Preview.
▪Through June 22, 2026, Fable 5 will be included in Pro, Max, and Team subscription plans at no extra cost before requiring usage credits.

Mythos-class model capabilities

▪Fable 5 can stay focused across millions of tokens in long-running tasks and improves its outputs using its own notes.
▪Anthropic states Claude Fable 5 shows exceptional performance in software engineering, knowledge work, vision, and scientific research.
▪On some benchmarks, Claude Fable 5 scored more than 10% higher than the company's previous model, Claude Opus 4.8.

Tiered safeguard system

▪The public release of Claude Fable 5 was made possible by new safeguards that block responses in high-risk areas.
▪Claude Mythos 5 is the same underlying model as Fable 5 but with safeguards lifted in some areas for trusted partners in Project Glasswing.
▪When a query triggers the safeguards, Fable 5 falls back to the less-capable Claude Opus 4.8 model to provide a response.
▪In early testing, more than 95% of Fable 5 sessions ran entirely on the model's own responses without any fallback.
▪Anthropic will require a 30-day data retention policy on all traffic for the new models to defend against novel attacks and jailbreaks.

Software engineering performance

▪AI platform Genspark reported that Fable 5 significantly outperformed other models on tasks like UI design and game coding.
▪Claude Fable 5 scored highest among frontier models on Cognition’s FrontierCode evaluation, which tests difficult coding tasks.
▪In early testing, Stripe reported that Claude Fable 5 performed a codebase-wide migration in one day that would have taken a team over two months.

Drug design applications

▪Using Claude Mythos 5, Anthropic's internal experts accelerated aspects of the drug design process by approximately ten times.
▪In one test, Mythos 5, using protein design tools without human assistance, was able to match or outperform skilled human operators.

Customer feedback

▪Financial firm IMC reported that Fable 5 "aced" its trading-analysis evaluations, including conceptual reasoning and root-cause analysis.
▪Analytics company Hex stated that Claude Fable 5 was the first model to score 90% on its core benchmark for complex analytical tasks.