Anthropic

The AI safety company behind the Claude model family, Project Glasswing, and the Mythos-class tier. Filed confidentially for an IPO in June 2026.

CEO and co-founder of Anthropic. Led the company through the Mythos program, Project Glasswing, the Fable 5 release, and the 2026 IPO filing.

Anthropic's first Mythos-class model as a product: $10 per million input tokens, $50 output, 1M context. The premium tier for long-horizon work.

Days-long autonomous runs, sub-agent delegation, persistent memory: what Fable 5 changes for agent builders, and where rivals stand today.

Model IDs, the 1M context window, refusal handling, fallbacks, and where Fable 5 runs today, from Copilot to Bedrock. A working setup guide.

Fable 5 ships with classifiers that reroute risky queries to Opus 4.8. Here is what gets blocked, what does not, and why Anthropic did it.

Fable 5 costs $10 per million input tokens and $50 for output. Here is who gets it, what the June 22 window means, and when to pick it.

A leaked draft, a security program with NATO on the partner list, and an IPO filing: the road to Claude Fable 5 was anything but quiet.

Fable 5 posts 80.3% on SWE-Bench Pro against 69.2 for Opus 4.8 and 58.6 for GPT-5.5. We ran the price-per-point math so you do not have to.

Anthropic just shipped its first Mythos-class model to the public. Here is what Fable 5 can do, what it costs, and why it is a new tier.