# anthropic-claude-fable-5-mythos-5-2026-06-09

## Veille

Anthropic launches Claude Fable 5 (a Mythos-class model made safe for general use) and Claude Mythos 5 (the same model, with guardrails lifted, restricted to cyberdefenders via Project Glasswing): state-of-the-art performance in software engineering, vision, long-context memory, and life sciences.

## Titre Article

Claude Fable 5 and Claude Mythos 5

## Date

2026-06-09

## URL

https://www.anthropic.com/news/claude-fable-5-mythos-5

## Keywords

Claude Fable 5, Claude Mythos 5, foundation model, Mythos class, autonomous agents, software engineering, computer vision, persistent memory, life sciences, protein design, cybersecurity, Project Glasswing, guardrails, alignment, token pricing, Opus 4.8

## Authors

Anthropic

## Ton

Profile: institutional corporate communication from an AI lab (corporate "we" perspective), controlled technical-promotional register, target audience of developers, researchers, and technical decision-makers. The style balances product announcement (benchmark records, customer testimonials) with a responsibility narrative (guardrails, classifiers, system card, data retention policy). Authority rests on quantified results, partner evaluations (Cognition, Hebbia, GitHub, Cursor), and concrete use cases (Stripe migration, protein design). Recurring metaphor of "the model working autonomously longer than ever before." Anthropic foregrounds an explicit dual objective: deploy fast AND safely, accepting deliberately conservative guardrails (false positives in under 5% of sessions).

## Pense-betes

- **Two models, one shared base**: Fable 5 = general-public version with guardrails; Mythos 5 = same model, guardrails lifted, restricted to cyberdefenders/infrastructure providers via Project Glasswing (collaboration with the US government).
- **Pricing**: $10/M input tokens, $50/M output tokens — less than half the price of Claude Mythos Preview.
- **Availability**: Fable 5 immediately on the Claude API and consumption-based plans; rollout to subscription plans June 9–22, 2026; Mythos 5 restricted to Glasswing partners + biology researchers (trusted access program forthcoming).
- **Three safety classifiers**: cybersecurity (0 compliance out of 30 jailbreak techniques in external testing), biology/chemistry (falls back to Opus 4.8), anti-distillation (unauthorized replication).
- **New policy**: 30-day data retention for Mythos-class models (safety monitoring).
- **Engineering**: Stripe completed a 50M-line Ruby codebase migration in one day (versus two months manually); highest score among frontier models on FrontierCode (Cognition), even at medium effort.
- **Vision**: state of the art; reconstructs a web app's code from screenshots; completes Pokémon FireRed with a vision-only harness.
- **Long-context memory**: focus on millions of tokens; persistent file-based memory improves performance 3× more than for Opus 4.8 (Slay the Spire).
- **Life sciences** (Mythos 5): protein design ~10× faster; molecular biology hypotheses preferred ~80% of the time in blind comparison against Opus-class models; genomics model 100× smaller outperforming a recent Science publication.
- **Conservative guardrails**: trigger in under 5% of sessions on average, with accepted false positives.

## RésuméDe400mots

On June 9, 2026, Anthropic announced the simultaneous launch of two models. **Claude Fable 5** is a "Mythos-class" model made safe for general use: its capabilities exceed those of any model Anthropic has publicly released, reaching state-of-the-art performance on nearly all benchmarks tested. **Claude Mythos 5** is the same underlying model, but with guardrails lifted in certain domains; it is restricted to a small group of cyberdefenders and infrastructure providers, deployed initially through Project Glasswing (in collaboration with the US government) as an upgrade to Claude Mythos Preview. Mythos 5 has the strongest cybersecurity capabilities of any model in the world.

Both models are priced at $10 per million input tokens and $50 per million output tokens, less than half the price of Mythos Preview. To deploy quickly and safely, Fable 5 ships with deliberately conservative guardrails (classifiers): on certain topics, the query receives Opus 4.8's response instead. On average, they trigger in fewer than 5% of sessions.

On capabilities, **software engineering**: Stripe reports that Fable 5 "compressed months of engineering into days," completing a 50-million-line Ruby codebase migration in one day (versus two months for a team). The model achieves the highest score among frontier models on FrontierCode (Cognition). **Knowledge work**: highest score of any model on Hebbia's Finance Benchmark (senior-level reasoning). **Vision**: state of the art; reconstructs a web app's source code from screenshots, completes Pokémon FireRed using vision alone. **Memory**: persistent file-based memory improves its performance 3× more than for Opus 4.8.

**Life sciences**: with Mythos 5, Anthropic's protein design experts accelerated the process roughly 10×; 9 of 14 protein targets produced strong candidates. Mythos 5 is the first model to generate novel, compelling scientific hypotheses, preferred ~80% of the time in blind comparison; it also conducted autonomous genomics research, training a model 100× smaller that outperforms a recent publication in Science. Automated alignment evaluation places Mythos 5's misaligned behavior at a low level, similar to Opus 4.8. Customer testimonials (Cursor, GitHub, Vercel, EvolutionaryScale) confirm autonomy on long-horizon tasks and reasoning superior to Opus 4.8.

## GrapheDeConnaissance

- Anthropic —a_créé→ Claude Fable 5 (TECHNOLOGIE, 0.99)
- Anthropic —a_créé→ Claude Mythos 5 (TECHNOLOGIE, 0.99)
- Anthropic —publie→ Claude Fable 5 (TECHNOLOGIE, 0.98)
- Claude Fable 5 —est_variante_de→ Claude Mythos 5 (TECHNOLOGIE, 0.95)
- Claude Fable 5 —est_instance_de→ classe Mythos (CONCEPT, 0.95)
- Claude Fable 5 —utilise→ classifieurs de sécurité (TECHNOLOGIE, 0.93)
- Claude Fable 5 —utilise→ Opus 4.8 (TECHNOLOGIE, 0.9)
- Claude Fable 5 —surpasse→ Opus 4.8 (TECHNOLOGIE, 0.9)
- Claude Mythos 5 —remplace→ Claude Mythos Preview (TECHNOLOGIE, 0.92)
- Claude Mythos 5 —fait_partie_de→ Project Glasswing (EVENEMENT, 0.9)
- Anthropic —collabore_avec→ gouvernement américain (ORGANISATION, 0.88)
- Claude Fable 5 —mesure→ meilleur score des modèles frontière sur FrontierCode (MESURE, 0.9)
- Claude Fable 5 —mesure→ 10 $/M tokens entrée et 50 $/M tokens sortie (MESURE, 0.95)
- Stripe —affirme_que→ Fable 5 a migré un codebase Ruby de 50M lignes en un jour (AFFIRMATION, 0.92)
- Claude Mythos 5 —améliore→ design de protéines accéléré ~10× (MESURE, 0.88)
- Claude Mythos 5 —surpasse→ modèles spécialisés de design de protéines (TECHNOLOGIE, 0.85)
- Claude Fable 5 —permet→ agents autonomes sur tâches à long horizon (CONCEPT, 0.9)
- Anthropic —recommande→ classe Mythos (CONCEPT, 0.85)

---
Canonical: https://www.thekb.eu/en/fiches/anthropic-claude-fable-5-mythos-5-2026-06-09/