Anthropic Launches Claude Fable 5 & Mythos 5: The New Mythos-Class Frontier

VEKA

VEKA

June 11, 2026 (Updated: June 12, 2026) · 3 min read

Anthropic Launches Claude Fable 5 & Mythos 5: The New Mythos-Class Frontier

One Model, Two Products

On June 9, 2026, Anthropic did something it has never done before: it shipped a single frontier model as two distinct products. Claude Fable 5 (API ID: claude-fable-5) is the new generally available flagship, wrapped in safety classifiers that hand certain sensitive queries off to Claude Opus 4.8. Claude Mythos 5 is the same underlying model with those safeguards lifted in some areas — and it is not generally available.

A New Tier Above Opus

Both models belong to a new "Mythos class" tier that sits above the existing Opus / Sonnet / Haiku hierarchy. The naming is deliberate: Fable (Latin fabula — "what is told," i.e. publishable) and Mythos (Greek — "myth, archetype," the deeper, less constrained form).

Benchmarks: A Significant Leap

Independent third-party testing shows Fable 5 pulling ahead of Opus 4.8 across the board:

  • SWE-Bench Verified: 95.0% vs 88.6% on Opus 4.8
  • SWE-Bench Pro: 80.3% vs 69.2% on Opus 4.8
  • FrontierCode Diamond: 29.3% vs 13.4% on Opus 4.8
  • Hex Core Analysis Benchmark: Hit 90% for the first time ever

In a real-world demonstration, Stripe migrated over 50 million lines of Ruby code in a single day using Fable 5's capabilities.

Pricing & Availability

The Mythos class is priced at $10 input / $50 output per 1M tokens — roughly double the cost of Opus 4.8. For a limited introductory window from June 9 through June 22, 2026, Fable 5 is included free for Claude Pro, Max, Team, and Enterprise users. After June 23, usage transitions to token-based billing.

Fable 5 is available via the Claude API, Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.

Safety: The Opus 4.8 Fallback

Fable 5 ships with the most aggressive safety scaffolding Anthropic has placed on a general release. If a request is flagged as related to cybersecurity exploitation, bio-chemical research, or model distillation attempts, the response is automatically routed to Claude Opus 4.8 instead. Anthropic says this fallback fires on fewer than 5% of sessions — over 95% of requests go through normally.

Mythos 5 & Project Glasswing

Claude Mythos 5 is the restricted sibling — same base model, some safeguards loosened. Access is limited to Project Glasswing-vetted partners, including infrastructure providers and approved cybersecurity researchers. In May 2026, Anthropic reported that roughly 50 partners had used Claude Mythos Preview to discover over 10,000 high- or critical-severity vulnerabilities. The program has since expanded to about 150 organizations across 15+ countries.

Key partners with Mythos 5 access include AWS, Microsoft, Apple, and CrowdStrike.

The Bigger Picture: Anthropic's $965B IPO

The release comes just days after Anthropic confidentially filed a draft S-1 with the SEC for an IPO, valuing the company at $965 billion following a $65 billion funding round. Revenue hit approximately $47 billion as of May 2026 and is expected to exceed $50 billion by end of June — a staggering leap from a $4 billion run rate in July 2025. OpenAI also filed for a listing on June 8, signaling a maturation phase across the AI sector.

What This Means for Developers

If you're building with Claude, Fable 5 is the model to target. It's strongest on longer, more complex tasks: software engineering, knowledge work, document analysis, vision, and multi-stage agent workflows. Paired with Claude Code, it functions closer to a senior project assistant than a simple drafting chatbot. The safety fallback means you can use it broadly without worrying about accidentally crossing restricted boundaries — the system handles that for you.

For teams in cybersecurity or biomedical research who need maximum capability, the Mythos 5 / Project Glasswing path is the way forward — but it requires vetting and approval.

VEKA

VEKA

Author at VEKA

Share this article

// THREAD

Discussion