Claude Mythos is 📝Anthropic's Mythos-class tier within the 📝Claude model family — its most capable system, positioned above Claude Opus 4.8 with major gains in reasoning, software engineering, agentic work, and cybersecurity. It reached the public on June 9, 2026 as 📝Claude Fable 5 and 📝Claude Mythos 5.

Claude Mythos began as an unreleased, limited-access 📝Large Language Model (LLM) — Claude Mythos Preview — that Anthropic considered too capable for public release and deployed through its Project Glasswing security initiative. On June 9, 2026 Anthropic released the Mythos-class model publicly in two forms: the safety-locked Claude Fable 5 and the less-restricted Claude Mythos 5. The naming departs from Anthropic's musical-term convention of Haiku, Sonnet, and Opus, marking a new top tier in the Claude lineup.

Key Facts

Category: Frontier LLM, Claude family (Mythos-class)
Developer: Anthropic
Tier positioning: Most capable Claude tier, above Claude Opus 4.8
Public release: June 9, 2026 — Claude Fable 5 (general availability) and Claude Mythos 5 (restricted)
Fable 5 model ID / pricing: claude-fable-5; $10 / input MTok, $50 / output MTok
Reported strengths: software engineering, multi-step reasoning, agentic workflows, cybersecurity
Context window: 1M tokens, 128K max output — see 📝Claude Mythos Context Window

Source: Anthropic — Claude Fable 5 and Claude Mythos 5.

How It Works

Claude Mythos extends the Claude architecture along three axes: deeper reasoning, longer autonomous action chains, and broader tool use.

Extended reasoning — improved multi-step planning with sustained coherence across long contexts.
Software engineering — state-of-the-art results on code generation, debugging, and refactoring.
Agentic capability — stronger tool use and goal-directed task completion through 📝Model Context Protocol (MCP).
Cybersecurity — Anthropic reports the Mythos Preview identified thousands of zero-day vulnerabilities across major operating systems and browsers, which is why the public Fable 5 ships with safety routing.

Why It Matters

Claude Mythos sits at the center of frontier-model governance: as capability rises, so does dual-use cyber and agentic risk. Anthropic restricted the most capable form — Mythos Preview — to vetted defenders under Project Glasswing, and shipped the public Claude Fable 5 with automatic safety routing that sends high-risk cybersecurity, biology, and distillation queries to Claude Opus 4.8.

The name also carries trademark weight. 📝MythOS has been in continuous commercial use since 2018 as a knowledge-management and human-AI augmentation platform at mythos.one. The two share only a name; formal correspondence with Anthropic's counsel is ongoing. See 📝MythOS vs Claude Mythos.

FAQ

What is Claude Mythos?

Anthropic's Mythos-class model tier — its most capable, above Claude Opus 4.8 — released publicly on June 9, 2026 as Claude Fable 5 and Claude Mythos 5.

When was Claude Mythos released?

June 9, 2026, as Claude Fable 5 (general availability) and Claude Mythos 5 (restricted). The earlier Claude Mythos Preview was a restricted research model used in Project Glasswing.

How is Claude Mythos different from Claude Opus?

It is Anthropic's tier above Claude Opus 4.8, with stronger reasoning, software engineering, agentic, and cybersecurity performance. Its name breaks from the Haiku / Sonnet / Opus convention.

Is Claude Mythos the same as MythOS?

No. MythOS is an independent knowledge-management platform at mythos.one, in commercial use since 2018. Claude Mythos is an Anthropic model that shares a name but not a product category. See 📝MythOS vs Claude Mythos.

Does Claude Mythos support MCP?

Yes, consistent with the Claude family's MCP support. See 📝Model Context Protocol and Claude Mythos.

📝MythOS vs Claude Mythos — disambiguation
📝Claude Mythos Release Date and What We Know So Far — timeline
📝Claude Mythos Benchmarks — reported performance
📝Claude Mythos for Developers — implications for builders
📝Claude Mythos and AI Safety — capability-risk tradeoffs
📝Claude Code — a primary surface for Claude models
📝What Anthropic's Claude Mythos and My Divorce Have in Common

Subjective

We have built on the Claude family since early API releases, and a tier above Opus changes which categories of autonomous work a small team can take on. The capability that makes Mythos a formidable security researcher is the same reason Anthropic shipped Fable 5 locked down. I love Anthropic's products; I hope the trademark concerns we have raised with their counsel resolve amicably. Both things are true.