Anthropic Launches Claude 5 — Then Immediately Blocks Its Most Powerful Version

On June 9, 2026, Anthropic released Claude Fable 5 to the public across Claude.ai, the API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Azure Foundry — its most capable model to date.

The same release introduced Claude Mythos 5 in limited availability to approved enterprise customers via a program codenamed Project Glasswing. Mythos 5 represented a full architecture upgrade, not a parameter scale-up — described internally as a "generational capability leap" over Opus 4.8.

Then on June 12, 2026 — just three days after launch — Anthropic announced it had disabled all access to Mythos-class models to comply with a directive from the U.S. Department of Commerce. No further details were disclosed. Fable 5 remained available; Mythos 5 access was suspended indefinitely.

Meanwhile, significant milestones were also crossed:

• Claude Sonnet 4 and Opus 4 were officially deprecated on June 15, 2026 — API calls using those model IDs now fail
• The entire Claude 3 generation has also fully retired (Opus in Jan 2026, Haiku in Apr 2026)
• MCP crossed 97 million installs in March 2026, cementing its role as the backbone of agentic AI

The Mythos 5 shutdown is unlike anything that has happened in commercial AI before.

Governments are now actively shaping which AI models the public can access. This isn't a lab making a cautious business decision — it's a regulatory intervention. The precedent is significant: if U.S. authorities can force a company to pull a model from the market, that changes the risk calculus for every AI lab and every enterprise building on frontier models.

For developers, the practical implications are real:

• Don't build on the most bleeding-edge models if regulatory continuity matters to your product
• Fable 5 is still available and represents a massive upgrade over Claude 4
• The deprecation of Sonnet 4 and Opus 4 means any app still calling those endpoints is now broken — check your API logs