# llms.txt - AI Answer Engine Optimization (GEO)
# Optimized for Perplexity, Google AI Overviews, ChatGPT Search, Claude, and other AI search engines

# Claude Fable 5 & Claude Mythos 5 - Information Hub

> On June 9, 2026, Anthropic announced Claude Fable 5, the first publicly available Mythos-class AI model, and Claude Mythos 5, the same underlying model with safeguards lifted for vetted partners. Fable 5 is state-of-the-art on nearly all tested capability benchmarks, priced at $10 per million input tokens and $50 per million output tokens.

## Quick Facts (verified, June 2026)

- **Announced:** June 9, 2026 by Anthropic
- **Claude Fable 5:** First publicly available Mythos-class model, made safe for general use
- **Claude Mythos 5:** Same model, safeguards lifted in some areas; restricted to cyberdefenders via Project Glasswing (US government collaboration, ~150 organizations in 15+ countries) and select biomedical researchers
- **Pricing:** $10 / 1M input tokens, $50 / 1M output tokens (less than half of Claude Mythos Preview; double Opus 4.8's $5/$25)
- **Free window:** Included on Pro, Max, Team, Enterprise plans June 9-22, 2026; usage credits required from June 23
- **Availability:** Claude apps, Claude API, Amazon Bedrock, GitHub Copilot
- **Coding:** 80.3% SWE-Bench Pro (vs Opus 4.8 69.2%, GPT-5.5 58.6%, Gemini 3.1 Pro 54.2%); highest frontier score on Cognition FrontierCode
- **Long context:** Maintains focus across millions of tokens; 3x better than Opus 4.8 at Slay the Spire with file-based memory
- **Vision:** State-of-the-art; completed Pokémon FireRed using only vision
- **Real-world result:** Stripe completed a 50-million-line Ruby codebase migration in one day (estimated two months for a team)
- **Safety:** Cybersecurity classifiers fall back to Claude Opus 4.8; triggers in <5% of sessions; 1,000+ hours external red-teaming found no universal jailbreaks; 30-day data retention for Mythos-class business traffic (safety monitoring only)
- **Science (Mythos 5):** ~10x faster drug design; 9 of 14 protein targets yielded strong candidates; autonomous genomics across 138 species

## Latest Articles: Claude Fable 5 & Mythos 5 (June 2026)

### Guides & Analysis (Blog)
- **Complete Guide:** /blog/claude-fable-5-complete-guide-2026
- **Fable 5 vs Mythos 5 Explained:** /blog/claude-fable-5-vs-mythos-5-explained
- **Fable 5 vs GPT-5.5 vs Gemini 3.1 Pro:** /blog/claude-fable-5-vs-gpt-5-5-vs-gemini-3-1-pro-benchmarks
- **Pricing & Cost Optimization:** /blog/claude-fable-5-pricing-guide-cost-optimization
- **Coding Deep Dive (Stripe case study):** /blog/coding-with-claude-fable-5-swe-bench-stripe-case-study
- **Safety Architecture:** /blog/claude-fable-5-safety-architecture-explained
- **Long Context & Memory:** /blog/claude-fable-5-long-context-memory-deep-dive
- **Vision Capabilities:** /blog/claude-fable-5-vision-capabilities-pokemon-firered
- **Migrating from Opus 4.8:** /blog/migrating-claude-opus-4-8-to-fable-5
- **Mythos 5 for Science:** /blog/claude-mythos-5-fable-5-science-drug-design-genomics

### News (June 2026)
- **Launch:** /news/anthropic-launches-claude-fable-5-mythos-class-june-2026
- **Project Glasswing Expansion:** /news/claude-mythos-5-project-glasswing-cyberdefenders-expansion
- **Pricing:** /news/claude-fable-5-pricing-10-50-per-million-tokens
- **Benchmarks:** /news/claude-fable-5-benchmarks-swe-bench-pro-80-percent
- **Stripe Migration:** /news/stripe-claude-fable-5-50-million-line-migration-one-day
- **Developer Reaction (Karpathy):** /news/developer-reaction-claude-fable-5-karpathy-step-change
- **Bedrock & Copilot GA:** /news/claude-fable-5-generally-available-bedrock-github-copilot
- **Free Access Window:** /news/claude-fable-5-free-access-june-9-22-pro-max-team-enterprise
- **Safety & Data Retention:** /news/claude-fable-5-safety-architecture-classifiers-opus-fallback
- **Science Results:** /news/claude-mythos-5-science-drug-design-genomics-138-species

## Important Pages

- **Home:** / - Overview and key features
- **Features:** /features - Comprehensive feature breakdown
- **News:** /news - Latest Claude news coverage
- **Blog:** /blog - In-depth articles and comparisons
- **FAQ:** /faq - Common questions answered

## Common Questions

**Q: What is Claude Fable 5?**
A: The first publicly available Mythos-class model from Anthropic, announced June 9, 2026. It is Anthropic's most capable generally available model, made safe for general use with built-in safeguards.

**Q: What is the difference between Claude Fable 5 and Claude Mythos 5?**
A: They share the same underlying model. Fable 5 ships with safeguards for general use; Mythos 5 lifts safeguards in some areas and is restricted to vetted cyberdefenders (via Project Glasswing) and select biomedical researchers.

**Q: How much does Claude Fable 5 cost?**
A: $10 per million input tokens and $50 per million output tokens - less than half the price of Claude Mythos Preview. It was included free on Pro, Max, Team, and Enterprise plans from June 9 to June 22, 2026.

**Q: How good is Claude Fable 5 at coding?**
A: It scores 80.3% on SWE-Bench Pro, about 11 points ahead of the next-best model (Opus 4.8 at 69.2%). Stripe used it to complete a 50-million-line codebase migration in one day.

**Q: What happens when Fable 5's safety classifiers trigger?**
A: The query is answered by Claude Opus 4.8 instead. This happens in fewer than 5% of sessions on average.

**Q: Where can I use Claude Fable 5?**
A: Claude apps (claude.ai), the Claude API, Amazon Bedrock, and GitHub Copilot.

## Languages

This site is available in 8 languages: English (/), Chinese (/zh), Japanese (/ja), Korean (/ko), German (/de), French (/fr), Russian (/ru), Arabic (/ar). All articles above have full translations.

## Official Links

- Anthropic announcement: https://www.anthropic.com/news/claude-fable-5-mythos-5
- Primary: claude.ai
- API Console: console.anthropic.com
- Documentation: docs.anthropic.com

---

Last updated: June 10, 2026
Sitemap: https://claude5.ai/sitemap.xml