Anthropic’s Mythos 1 Is Coming to Claude Code and Claude Security — Here’s What We Know

Just days after the claude-mythos-1-preview model string briefly surfaced inside Claude’s UI, Anthropic appears to be making its next move. New internal strings and official retailer-style listings confirm that Mythos 1 — the successor to the restricted Claude Mythos Preview — is being actively prepared for deployment inside Claude Code and Claude Security. And if the image doing the rounds is anything to go by, the rollout may be closer than anyone expected.

What Just Happened?

The model string claude-mythos-1-preview appeared — briefly — inside the Claude interface for a short window before being pulled. Simultaneously, new internal strings referencing Mythos have been added across Anthropic’s platform infrastructure. The access scope described reads clearly:

“Access to the Claude Mythos model in Claude Code and Claude Security.”

This is not a general public release. But it is a significant step beyond where Anthropic stood just six weeks ago.

Mythos: A Quick Recap

Mythos Preview demonstrated a striking leap across evaluation benchmarks compared to Claude Opus 4.6 — scoring 93.9% on SWE-bench Verified versus 80.8%, and 97.6% on USAMO 2026 versus 42.3%. Despite those numbers, Anthropic deliberately chose not to release it publicly, citing its unprecedented capabilities in autonomously discovering software vulnerabilities.

Why Claude Code and Claude Security?

Claude Mythos Preview marked a major inflection point in how software vulnerabilities are discovered and exploited — shifting the speed, scale, and sophistication of vulnerability discovery to a level security teams can no longer treat as incremental.

Deploying Mythos 1 inside Claude Code makes immediate sense — the model’s agentic coding capabilities are, by benchmarks, the most advanced ever measured. Developers using Claude Code for security-sensitive projects would get direct access to a model that can autonomously find and reason about critical code flaws.

Claude Security, meanwhile, is getting structural upgrades alongside this rollout. A new dashboard is being built that surfaces discovered vulnerabilities, seven-day and thirty-day historical charts, and deeper triage results — bringing it closer to parity with how rival security suites present scan history.

Also Read: Claude Code: What It Is and How to Use It

Anthropic’s Shifting Stance on Public Access

When Mythos Preview launched in April, Anthropic was categorical: this model would not be publicly available. Six weeks and ten thousand-plus vulnerabilities later, the language has shifted.

A recent Project Glasswing update notes that the model is now helping protect a wider range of organizations, including open-source projects, and adds that Mythos-grade models could reach the public once the right safeguards are in place — a notable shift from the earlier framing.

Anthropic’s own statement on May 22 put it plainly: “In the near future, once we’ve developed the far stronger safeguards we need, we look forward to making Mythos-class models available through a general release.”

That’s a door opening — slowly, carefully, but opening.

Another signal: by May 17, the “Preview” label that had accompanied every Mythos listing since April disappeared from Anthropic’s internal model listing on Google Cloud — the same sequence that preceded the broader availability of Claude Opus 4.7 on Vertex AI.

Also Read: Anthropic Claude Models Explained: Opus vs Sonnet vs Haiku

What Mythos 1 Is NOT (Yet)

Let’s be clear about what this development means — and doesn’t mean:

• Not a public release. Access remains gated to Claude Code and Claude Security users — primarily enterprise customers.
• Not a replacement for Opus 4.7 or Sonnet 4.6 for general chat use.
• Not confirmed for API access for independent developers — yet.

There are no indications yet that the product will move beyond enterprise customers, but the direction of travel is unmistakable.

The Bigger Picture

Claude Mythos Preview was announced after an accidental leak on March 26, subsequently prompted the Trump administration to consider government pre-deployment oversight of frontier AI models, and found a 27-year-old bug in OpenBSD that survived decades of expert review. It is, by any measure, a category-defining model.

The fact that Anthropic is now funnelling it into its developer and security products — even in a controlled way — signals that the company believes it has sufficient guardrails in place for these specific, monitored use cases. Mythos 1 in Claude Code could give developers the most capable coding AI ever shipped in a product. Mythos 1 in Claude Security could fundamentally change how enterprises approach vulnerability management.

The era of Mythos-class AI being strictly lab-only appears to be ending. Quietly, methodically — but ending.

Stay tuned to TechnoSports for the latest in AI model developments, Anthropic news, and enterprise tech coverage.