Anthropic appears to be moving its restricted Claude Mythos model toward a public release, with the powerful system surfacing inside Claude Code and Claude Security before vanishing again.
Key Points:
- References to a model called Mythos 1 briefly appeared in public versions of Claude Code and Claude Security.
- The model is labeled claude-mythos-1-preview, signaling preparation for a wider rollout.
- Anthropic had kept Mythos restricted after it found more than 10,000 critical software vulnerabilities.
Mythos 1 Nears Public Rollout
Anthropic looks ready to widen access to Claude Mythos, the most powerful AI model it has built so far. Reports from BleepingComputer and Cyber Security News this week point to a staged commercial release tied to its developer and security products.
Some users briefly spotted a toggle to enable Mythos in the public version of Claude Code before it was pulled offline. New code strings reference the model as claude-mythos-1-preview, alongside a line describing access to Mythos within Claude Code and Claude Security.
The company first announced Mythos on Apr. 7 as an early-preview frontier model with striking strength in computer security tasks.
Anthropic had warned that the model carried severe risks to global digital infrastructure. It said Mythos could automatically build functional cyberattacks at a professional level, which is why a public rollout was held back until stronger guardrails were ready.
Also Read: Bitcoin Volatility Sinks To An 8-Month Low As Bears Crowd Resistance
Glasswing Reveals Mythos Risks
Rather than ship Mythos broadly, Anthropic launched Project Glasswing, a defensive initiative giving roughly 50 partner organizations exclusive access to the model.
Those partners used the preview to scan widely deployed open-source software. They uncovered more than 10,000 high or critical-severity vulnerabilities within the first month, a haul that explains the company's caution.
That scale is why the Claude Code sighting matters. Anthropic itself wrote that "the advantage will belong to the side that can get the most out of these tools," framing the technology as a race between attackers and defenders.
Anthropic argues defenders will win over time, fixing bugs before flawed code ever ships. Critics counter that the same reasoning gains shorten the path from finding a bug to weaponizing it, and verification still moves slower than discovery.
Mythos Joins a Crowded Lineup
Mythos arrived after a turbulent debut. The model first surfaced in late Mar. when a content management misconfiguration exposed nearly 3,000 internal files, including a draft post describing it as Anthropic's most powerful system.
Anthropic confirmed the model days later and now offers Claude Opus 4.7, Opus 4.6, Opus 4.5, Sonnet 4.6, and Haiku 5.5, with Mythos positioned as a separate, higher tier still gated behind enterprise access.
Read Next: Ethereum Price Slips Below $2,100 As Buying Demand Quietly Cools