📝Claude Mythos is not publicly available. As of April 2026, 📝Anthropic has stated they do not plan to make Claude Mythos Preview generally available — access is restricted to vetted organizations through 📝Project Glasswing, their invite-only program for defensive cybersecurity applications.
Who Can Access Claude Mythos
Project Glasswing is limited to approximately 50 organizations — 12 founding partners and ~40 additional vetted critical-infrastructure organizations. Founding partners include 📝Amazon, 📝Apple, 📝Google, 📝JPMorgan Chase, 📝Microsoft, 📝NVIDIA, Broadcom, 📝Cisco, 📝CrowdStrike, the Linux Foundation, and Palo Alto Networks.
Approved participants access the model through the Claude API, Amazon Bedrock, Google Cloud's Vertex AI, and Microsoft Foundry at $25/$125 per million input/output tokens.
How to Apply
Applications are accepted at anthropic.com/glasswing. Eligibility is concentrated on organizations that build or maintain critical software infrastructure. There is no self-serve sign-up.
What You Can Use Now
The most capable publicly available Claude model is Claude Opus 4.6 — accessible through claude.ai, the Claude API, 📝Claude Code, Amazon Bedrock, and Vertex AI. Opus 4.6 is fully supported in third-party clients like 📝Cursor and IDE extensions.
For identity-aware AI that knows your context, 📝MythOS (different from Claude Mythos) connects to any Claude-powered client via the 📝Model Context Protocol (MCP) — giving agents access to your knowledge library, memos, and perspectives regardless of which model you're running.
When Will Claude Mythos Be Public?
Anthropic has indicated they plan to bring "Mythos-class capabilities to a future Claude Opus release with additional safety safeguards." Industry estimates suggest Q3 2026, though no official date has been confirmed. See 📝Claude Mythos Release Date and What We Know So Far for the latest.
Platform-Specific Guides
- 📝How to Use Claude Mythos with Cursor — Cursor-specific access and workarounds
