Anthropic released Claude Fable 5 on June 9, the first publicly available model in its new Mythos-class tier. It sits above the existing Opus line, and early testing suggests it’s a significant step up in coding and reasoning. The release landed the same day as WWDC, making for a busy Tuesday in tech.
What It Can Do
The most striking demos so far involve code generation. AI researcher Ethan Mollick, an associate professor at Wharton, described the model as consistently outperforming every other public model he’s used by a considerable margin. His tests included generating fully playable 3D browser games from a single sentence: a Library of Babel explorer, a self-aware version of Snake, and a complete run of Slay the Spire played autonomously by the model. The games render in 3D and run in a browser tab.
Anthropic says Fable 5 targets what it calls long-horizon coding: complex, multi-step software engineering tasks that previous models would stumble on or abandon partway through. Independent benchmarks put it at roughly 80% on SWE-Bench Pro, which measures real-world software engineering performance.
Safety Guardrails
The release comes with hard limits. In high-risk areas including cybersecurity, biology, chemistry, and distillation, the model blocks responses and falls back to Claude Opus 4.8. Anthropic has been clear about why: Fable 5 is a consumer-facing version of a model that was previously restricted to vetted infrastructure providers and cybersecurity researchers due to capability concerns. The public version ships with those concerns built into the guardrails from day one.
Pricing and Availability
Fable 5 is available now through the Claude API, Amazon Bedrock, and GitHub Copilot. The API model ID is claude-fable-5. List pricing is $10 per million input tokens and $50 per million output tokens, twice the cost of Opus 4.8. The model supports up to 128,000 output tokens and has a knowledge cutoff of January 2026.
Claude Pro, Max, Team, and Enterprise subscribers get access at no extra charge through June 22. After that, usage draws on credits or usage-based billing depending on the plan.