Anthropic releases Claude Fable 5 at $10/$50 per million tokens, posting 80.3% on SWE-Bench Pro vs GPT-5.5's 58.6%

Anthropic on June 9 made Claude Fable 5 generally available across its API, apps and subscription plans, the first public release from its new Mythos-class tier that sits above the Opus line. Fable 5 shares underlying weights with Claude Mythos 5, a safeguards-lifted variant restricted to Project Glasswing government cybersecurity partners and vetted biology researchers, with a 30-day data-retention policy for Mythos-class usage. Pricing is $10 per million input tokens and $50 per million output tokens, roughly double Claude Opus 4.8.

Anthropic published benchmarks showing state-of-the-art results on nearly all tested suites, including 80.3% on SWE-Bench Pro versus 58.6% for OpenAI's GPT-5.5, the top FrontierCode score among frontier models at medium effort, and the first result above 90% on its core analytics benchmark — a 10-point jump over Opus. Misuse classifiers covering cybersecurity, biology-chemistry and distillation route flagged sessions to Opus 4.8; the company says more than 95% of sessions trigger no fallback and that over 1,000 hours of external red-teaming produced no universal jailbreaks. Stripe reported Fable 5 completed a 50-million-line Ruby codebase migration in one day, an estimated two months of manual work.

The model went live same day on GitHub Copilot and Amazon Bedrock and is included in Pro, Max, Team and Enterprise plans at no extra cost through June 22.