Anthropic just dropped Claude Fable 5, the public-facing version of its most capable AI system to date.
The model launched on June 9 as a dual release. The mainstream Fable 5 comes loaded with safety measures for general use, while its more powerful sibling, Mythos 5, remains locked away for select partners working on something called Project Glasswing.
What Fable 5 actually does, and what it won’t
Fable 5 is designed to handle software engineering, knowledge work, scientific research, and complex vision tasks. When it encounters high-risk inquiries in areas like cybersecurity, biology, or chemistry, it redirects those queries to the previous model, Claude Opus 4.8, which handles them under tighter constraints. This fallback mechanism kicks in for less than 5% of total use cases.
Access is broad and immediate. Users on paid Claude plans can start using Fable 5 right now. It’s also available through APIs, cloud services from Amazon, Google, and Microsoft, and through GitHub Copilot integration. During the introductory period, Fable 5 is included at no extra cost on paid plans until June 22.
After that grace period, API pricing is set at $10 per million input tokens and $50 per million output tokens.
The Mythos 5 question
Mythos 5, the unrestricted variant, is available only to select partners involved in Project Glasswing and specific biology researchers.
The launch follows a cautious development period that included limited previews earlier in 2026. Anthropic’s Responsible Scaling Policy has guided the approach, with internal assessments identifying significant risks associated with advanced AI applications in cybersecurity and biosafety driving the decision to restrict the most capable variant.
What this means for the tech landscape
For developers and companies already embedded in the AI tooling ecosystem, the GitHub Copilot integration is the most immediately relevant detail. Software teams using Copilot will gain access to Fable 5’s enhanced capabilities for code generation and engineering tasks.
The pricing structure — $10 per million input tokens and $50 per million output tokens — will be worth watching closely. The free introductory period through June 22 gets teams dependent on the capability before the meter starts running.
One thing conspicuously absent from the entire launch is any connection to crypto or digital assets. Anthropic’s silence on the subject is itself a signal, consistent with the company’s focus on enterprise AI and safety credibility.