Anthropic Opens Mythos-Class to the Public: Claude Fable 5 Goes Live for Everyone
Anthropic has released Claude Fable 5 and Claude Mythos 5 to the general public — marking the first time a Mythos-class model is available outside of restricted government and enterprise programs. The announcement, published Monday on Anthropic's blog, represents a major milestone in AI accessibility: the company's most capable model, previously gated behind Project Glasswing and trusted-access programs, is now available to any developer or user with an API key.
The move comes just weeks after Anthropic's confidential IPO filing at a $965 billion valuation, and positions Fable 5 as the company's flagship model heading into its public market debut. Pricing is set at $10 per million input tokens and $50 per million output tokens — less than half the cost of Claude Mythos Preview and slightly above Opus 4.8 rates.
What Makes Fable 5 Different
Claude Fable 5 is not an incremental update. Anthropic's own evaluations show it achieving state-of-the-art results across software engineering, finance, life sciences, and vision benchmarks. Customer testimonials in the announcement paint a striking picture: Cursor calls it "state of the art on CursorBench," GitHub says it "opens up a class of long-horizon problems," and Hebbia reports it's "a clear step beyond Opus 4.8" at senior research scientist grade.
Notable real-world results from the launch materials:
- Software engineering: Stripe reports Fable 5 compressed months of engineering into days — migrating a 50-million-line Ruby codebase in a single day that would have taken a team over two months manually.
- Vision: Fable 5 can rebuild a web app's source code from screenshots alone. It beat Pokémon FireRed using only raw visual input — no maps, no navigation aids — a task earlier models required extensive scaffolding to accomplish.
- Scientific research: Mythos 5 produced novel, compelling molecular biology hypotheses that scientists preferred over Opus-class models ~80% of the time in blinded comparisons. One hypothesis was later independently validated by a separate lab.
- Autonomous operation: Across multi-million-token contexts, Fable 5 stays focused and self-corrects using its own notes — improving performance 3x more than Opus 4.8 on complex game-playing benchmarks.
Safety at Scale
The release comes with a critical asterisk. Anthropic has acknowledged that Mythos-class models present "substantial risk of catastrophic misuse" — particularly in cybersecurity and research biology — and has deployed a new set of guardrails that include built-in classifiers to detect jailbreak attempts and filter sensitive topics. These classifiers are deliberately cautious; Anthropic warns that benign requests will sometimes trigger them, producing false positives the company aims to reduce post-launch.
"Because we have prioritized safety, we've deliberately tuned the safeguards to be cautious, and they are still stricter than would be ideal." — Anthropic
For cyberdefenders and critical infrastructure providers, Mythos 5 lifts some of these restrictions and is available through Project Glasswing and an expanding trusted-access program. Anthropic says it plans to "expand access over time through a more systematic trusted-access program" — a structure that creates a two-tier system: unrestricted access for vetted security professionals, and guarded access for general users.
Market Implications
The public release of Mythos-class capability shifts the competitive landscape significantly. Google and OpenAI have both filed for IPOs, and Google just slashed its AI Plus subscription price from $7.99 to $4.99 — firing a warning shot in the AI subscription price wars. Anthropic's Fable 5 launch raises the capability floor for what any paying user can access, putting pressure on competitors to match or differentiate.
On Hacker News, the announcement quickly became the #1 story with over 2,200 points — reflecting the developer community's recognition that this is not a routine model update. The availability of Mythos-class capability through a standard API means the boundaries of autonomous AI agents, long-horizon software engineering, and AI-driven scientific research just shifted for every company building on LLMs.
Sources: Anthropic Blog, The Verge, TechCrunch, Hacker News