Anthropic launches Claude Fable 5 and Claude Mythos 5
TL;DR
Anthropic released Claude Fable 5 and Claude Mythos 5 on 9 June 2026. Fable 5 is a Mythos-class frontier model with full safeguards at $10/$50 per million tokens; Mythos 5 lifts select safeguards for vetted cybersecurity and biology researchers.
What changed
Anthropic released Claude Fable 5 and Claude Mythos 5 on 9 June 2026, its next generation of frontier models for hard knowledge work and coding. Fable 5 is the generally available model: Mythos-class intelligence with full safeguards, priced at $10 per million input tokens and $50 per million output. Mythos 5 is the same underlying model with select safeguards lifted, restricted to Project Glasswing cybersecurity partners and vetted biology researchers.
Specs
- •Pricing: $10 per million input tokens and $50 per million output, less than half the Claude Mythos Preview rate
- •Availability: API from 9 June 2026; included on Pro, Max, Team, and seat-based Enterprise plans until 22 June, then via usage credits
- •Mythos 5 access: Project Glasswing partners plus vetted biology researchers, with a 30-day data retention policy
- •Safeguards: classifiers route sensitive cyber, bio, and distillation requests to Opus 4.8; no universal jailbreaks in over 1,000 hours of external red-teaming
Why it matters
Anthropic says the pair beats Opus 4.8 on nearly all tested benchmarks, with the biggest gains on long-horizon autonomous work. Stripe reported that Fable 5 compressed months of engineering into days on a 50-million-line codebase migration, and Cursor says it opened up a class of long-horizon problems earlier models could not reach. At less than half the price of Claude Mythos Preview, senior-engineer-grade autonomy just got materially cheaper.
What to watch for
Fable 5 is included on Pro, Max, Team, and seat-based Enterprise plans at no extra cost until 22 June; after that it draws usage credits until capacity improves. Watch how the safeguard routing behaves in practice: sensitive requests fall back to Opus 4.8 rather than being refused outright. Anthropic also reports no universal jailbreaks across more than 1,000 hours of external red-teaming, a claim the security community will now test in the open.
Who this matters for
- Developers get the strongest coding model Anthropic has shipped, with Stripe and GitHub citing month-to-day compression on large engineering tasks.
- Vibe Builders on Pro or Max plans can use Fable 5 at no extra cost until 22 June 2026, then via usage credits.
- Security and biology researchers can apply for Mythos 5 access through the Project Glasswing trusted access program.
Harsh’s take
The interesting move is productizing the safety split: one frontier brain, two doors. Fable 5 gives everyone the capability, with guardrails that route risky requests to Opus 4.8 instead of refusing them; Mythos 5 hands vetted researchers the raw version under a 30-day retention policy.
For builders the practical headline is price. $10 in and $50 out per million tokens is less than half the Mythos Preview rate, and the free window on paid plans until 22 June is the cheapest way to find out whether your hardest workflow is now automatable.
by Harsh Desai
About Claude
View the full Claude page →All Claude updatesMore AI news
- FeaturePackage Firewall: Blocking 8,000+ malicious packages daily
Replit launched Package Firewall in partnership with Socket. The feature blocks over 8,000 malicious packages daily at installation time.
- FeatureBenSyc: Benchmarking Sycophancy and Alignment in Bengali LLMs
BenSyc introduces a benchmark to measure conversational sycophancy and human alignment in LLMs for Bengali contexts.
- FeatureCodeAlchemy framework generates synthetic data for code model training
CodeAlchemy introduces a synthetic data generation framework to address limited signals from raw code pre-training for diverse real-world tasks.