Anthropic Unveils Claude Fable 5.2 and Mythos 5.2 With 20% Lower API Pricing
Sharing identical weights, Claude Fable 5.2 opens globally at $8 per million input tokens while vetted cyber and life-science groups get Mythos 5.2 with fewer guardrails.

By Jin, Founder & Startups
, updated 2 min read
Anthropic on November 9 rolled out Claude Fable 5.2 and Claude Mythos 5.2, cutting base frontier API pricing by 20% while pitching a dual-tier deployment model for open-ended engineering and restricted dual-use research.
Both releases run on identical underlying weights and retain a 1-million-token context window, Anthropic confirmed in its release notes. Claude Fable 5.2 ships with strict cybersecurity and biology safeguards across Claude Pro, Max, Ultra, Team, Enterprise, and major cloud APIs, whereas Claude Mythos 5.2 relaxes selected content blocks exclusively for vetted organisations inside Anthropic's Life Sciences and Cyber Verification Programs.
Faster Token Output and Reduced API Rates
Standard pricing: Drops from $10 input / $50 output on 5.1 to $8 per million input tokens and $40 per million output tokens ($0.20 per million cached read tokens), generating outputs up to 30% faster.
Optional fast mode: Increases generation throughput up to 2.5x at $16 per million input tokens and $80 per million output tokens.
Advisor and orchestrator roles: Anthropic highlighted workflows where Opus 5.5 handles day-to-day software tasks and calls Fable 5.2 selectively as an Advisor, or uses Fable 5.2 as an Orchestrator delegating steps to Haiku and Sonnet workers.
In pre-release engineering and science tests, Anthropic reported that the model solved two open black-hole physics problems, designed three functional electronics prototypes from scratch, lifted hardware test Diode PCB Gym scores to 69%, reached 83.5% on BenchCAD 2.1, and formalised the proof of Fermat's Last Theorem in a 30-hour agentic run.
Sources
- 1Anthropic: Introducing Claude Fable 5.2 and Mythos 5.2anthropic.com