Claude Fable 5: Specs, Pricing & Best Use Cases

Claude Fable 5 is Anthropic's first publicly available Mythos-class model, launched June 9, 2026 above the Opus tier. It's built for long-running, autonomous knowledge work and coding, with a 1M-token context window and state-of-the-art scores on nearly every benchmark Anthropic has published. This guide covers its specs, pricing, benchmarks, the export-control disruption it went through, and how it fits into Chat Smith's current model lineup.

Claude Fable 5 (model ID: claude-fable-5) sits in Anthropic's Mythos class, a tier the company positions above Opus. It's the generally available version of Mythos-level capability: Fable 5 and Claude Mythos 5 share the same underlying model, but Fable 5 ships with safety classifiers that redirect high-risk requests — cybersecurity, biology, chemistry — to a Claude Opus 4.8 fallback instead of answering directly. Mythos 5 has some of those safeguards lifted and stays restricted to a small group of vetted cybersecurity partners.

Anthropic says Fable 5 is state-of-the-art on nearly all tested capability benchmarks, with its lead over other models growing as tasks get longer and more complex. It's designed to plan across stages, delegate to sub-agents, write its own tests, and check its own work over multi-day autonomous sessions.

Three days after launch, on June 12, 2026, the U.S. government applied export controls to Fable 5 and Mythos 5, after Amazon researchers reported a way to bypass Fable 5's safeguards. Anthropic suspended access for all users worldwide rather than risk non-compliance. The Department of Commerce lifted the controls on June 30, and Anthropic restored global access on July 1 with an improved safety classifier that blocks the reported bypass technique.

Fable 5 is Anthropic's attempt to bring its most capable model to everyone without the risks of an unrestricted release. Real-time classifiers watch every session and automatically block or reroute requests touching cybersecurity, biology, chemistry, or weapons-relevant topics, falling back to Opus 4.8 for those specific queries while leaving everything else at full Mythos-class capability.

Fable 5 accepts text, images, documents, and structured data, all within a 1M-token context window billed at a single flat rate — a 900K-token request costs the same per token as a 9K-token one. That makes it practical for tasks like rebuilding a web app from screenshots or reasoning over an entire codebase in one pass.

Run in an agent harness like Claude Code, Fable 5 can work for days at a time: planning across stages, delegating to sub-agents, writing its own tests, and using vision to check outputs against goals. Anthropic reports Stripe compressed a two-month engineering task into a single day using the model.

Fable 5 costs twice as much as Opus 4.8 per token, and Anthropic's own safeguards route some of its queries to Opus 4.8 anyway. The premium pays off on the hardest end of the task spectrum — long, ambiguous, multi-step work — where Fable 5's lead over other models, including GPT-5.5, grows with task complexity. For everyday coding and writing, Opus 4.8 remains the better economics.

At double the cost of Opus 4.8, Fable 5 earns its price on work that's genuinely long, ambiguous, or high-stakes:

• Large codebase migrations. Multi-day autonomous coding sessions that plan, delegate to sub-agents, and write their own tests along the way.
• Rebuilding interfaces from screenshots. Vision input lets Fable 5 implement designs with high fidelity and check its own output against the original image.
• Senior-level financial and legal reasoning. Long documents, complex multi-step analysis, and high-stakes reviews where accuracy matters more than cost per call.
• Scientific research workflows. Complex, multi-step research tasks that benefit from the model's ability to recover from failures and keep working without frequent check-ins.
• Ambiguous, open-ended problems. Tasks that don't have a clear spec upfront benefit most from Fable 5's planning and self-correction, where its lead over smaller models is largest.

Claude Fable 5 isn't integrated into Chat Smith. The platform currently offers Claude Sonnet 5, Anthropic's balanced mid-tier model, alongside frontier models from OpenAI, Google, and xAI — all in one chat app.

• Run everyday coding and writing tasks on Claude Sonnet 4.6 or Sonnet 5
• Compare Claude's answers against Grok 4.5 or Gemini 2.5 Pro
• Switch models mid-conversation without losing your chat history

Claude Fable 5 is Anthropic's first attempt at bringing Mythos-class capability to the public, and it shows on the hardest, longest tasks — at double the cost of Opus 4.8 and with real-time safeguards that fall back to Opus 4.8 on flagged topics. It survived a brief export-control suspension in June 2026 and is back to full global availability. It's not currently on Chat Smith, but Claude Sonnet 5 covers most everyday use cases alongside other leading models in one app.