Breaking — Anthropic just launched Claude Fable 5, the first Mythos-class model available to the public. It is state-of-the-art on nearly every benchmark — and it comes with guardrails that didn’t exist six months ago.
Update, 26 Oct: Anthropic’s S-1 shows the filed numbers.Anthropic’s S-1 dates the launch: Fable 5 became generally available in June 2026, after Claude Mythos Preview was offered through Project Glasswing in April. With Fable 5.1 in September 2026, Anthropic says it reduced the cost of typical workloads by about 25% compared with Fable 5, with savings of up to 45% for complex coding and agentic tasks. Revenue was $11,555 million for the three months ended 30 June 2026. More: our S-1 piece.
Fable 5 is Mythos-class capability with a safety harness bolted on top:
High-risk domain blocking: Queries about cybersecurity exploitation, biological threats, or chemical synthesis automatically fall back to Claude Opus 4.8
Conservative tuning: Guardrails trigger in less than 5% of sessions — so 95%+ of users never hit them
Jailbreak-resistant: 1,000+ hours of internal bug bounty + external red-teaming found zero universal jailbreaks
Distillation blocking: Cannot be used to train competing models
This is the Harness Theory applied to safety: Anthropic didn’t make Mythos safe by making it weaker. They built a routing layer that sends dangerous queries to a less capable model while keeping everything else at Mythos-class performance.
Pricing: Aggressive
$10 input / $50 output per million tokens — less than half of Mythos Preview pricing. With 90% prompt caching discount, effective input cost drops to $1/million tokens.
Through June 22, Fable 5 is included in Pro, Max, Team, and Enterprise plans at no extra cost. After June 22, usage credits required.
The pricing signals intent: Anthropic wants Fable 5 to be the default model, not a premium tier. At $10/$50, it undercuts most frontier model — as explored in the intelligence factory race between AI labs — s while outperforming them on benchmarks.
The Map of AI Read
Fable 5 is a Layer 4 (Models) event with three structural implications:
The frontier just moved — again. 80.3% on SWE-Bench Pro vs OpenAI’s 58.6%. The gap between Anthropic and OpenAI on software engineering is now 22 percentage points. That matters because software engineering is where enterpriserevenue concentrates.
Safety as routing. Anthropic’s safeguard architecture — fallback to a weaker model for dangerous queries — is a prototype of what every frontier model will need. Apple’s three-tier routing for Siri AI does the same thing. Safety is becoming an orchestration problem, not a model problem.
Pricing compresses Layer 4 margins. At less than half of Mythos Preview pricing, Anthropic is actively commoditizing its own model layer — the same move Apple made by declaring models interchangeable. The Product Overhang in capability is being released as a price collapse.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
