Anthropic has released Claude Fable 5, the first publicly available model from its top-secret Mythos tier — a new class of AI that sits above the entire Opus lineup. Launched on June 9, 2026, Fable 5 is the biggest capability jump Anthropic has shipped to the public and is available today via the Claude API, Claude.ai subscription plans, Amazon Bedrock, Vertex AI, Microsoft Foundry, and GitHub Copilot.
What Is Claude Fable 5?
Fable 5 shares identical weights with Claude Mythos 5 — the restricted model Anthropic launched in April 2026 for a small number of cybersecurity and critical infrastructure partners under Project Glasswing. The difference: Fable 5 ships with safety classifiers that reroute high-risk queries in areas like cybersecurity, biology, and chemistry to Claude Opus 4.8 instead of answering directly. According to Anthropic, these classifiers fire in fewer than 5% of sessions.
Benchmark Results
Fable 5 leads every benchmark Anthropic has tested. On SWE-Bench Pro — which measures real software engineering task completion on GitHub repositories — it scores 80.3%, versus 69.2% for Claude Opus 4.8, 58.6% for GPT-5.5, and 54.2% for Gemini 3.1 Pro. On SWE-bench Verified it reaches 95%, up from 88.6% for Opus 4.8. Andrej Karpathy described it as “a major-version-bump-deserving step change forward.”
Key Specs
- Context window: 1 million tokens
- Max output: 128,000 tokens
- API model ID: claude-fable-5
- Pricing: $10 / $50 per million input/output tokens
- Free on subscription plans (Pro, Max, Team, Enterprise) through June 22, 2026; usage credits required after
Real-World Impact
Stripe tested Fable 5 on a 50-million-line Ruby codebase and completed a full migration in a single day — work a human engineering team would have needed over two months to finish. The 1 million token context window makes this possible by fitting entire codebases into a single request.
Read our full Claude Fable 5 review for a complete breakdown of benchmarks, pricing, access methods, and who should use it.