Anthropic launches Claude Fable 5 Mythos-class AI for public use

Anthropic’s decision to put a Mythos‑class model into general circulation marks a new phase in frontier AI deployment. Fable 5 is essentially the public twin of the Mythos architecture that previously lived behind a trusted‑access firewall, now wrapped in aggressive safety classifiers that auto‑route high‑risk cyber and bio requests to weaker models like Opus 4.8.([axios.com](https://www.axios.com/2026/06/09/anthropic-mythos-class-safeguards?utm_source=openai)) In practice, this gives enterprises and serious developers access to substantially more capable systems for coding, research, and long‑horizon agentic tasks without fully unleashing the scarier capabilities that spooked regulators and red‑teamers earlier in the year.([techcrunch.com](https://techcrunch.com/2026/06/09/anthropic-released-claude-fable-5-its-most-powerful-model-publicly-days-after-warning-ai-is-getting-too-dangerous/?utm_source=openai))

Strategically, Fable 5 tightens the duopoly emerging between Anthropic and OpenAI at the very top of the model stack. Both are now converging on a two‑tier approach: ultra‑capable models under restrictive programs for security‑sensitive work, and slightly neutered versions for the broader market. This structure lets them monetise frontier capability, gatekeep who gets offensive cyber tools, and position themselves as indispensable infrastructure for governments and large firms.([axios.com](https://www.axios.com/2026/06/09/anthropic-openai-mythos-ai-model-access?utm_source=openai)) It also raises the bar significantly for rivals: competing means not just matching raw capability, but also building comparable evaluation pipelines, guardrails, and access‑control regimes that regulators will trust.