Anthropic Launches New Claude Fable and Mythos Models
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on Monday. The company calls them its most advanced models.
- 30% faster outputs than the 5.1 generation, the company says.
- $8 per million input tokens for Fable, down from $10.
- $40 per million output tokens, a 20% cut from 5.1 pricing.
- Mythos 5.2 goes only to vetted organizations through verification programs.
Why it matters: This is Anthropic's first frontier release since Amodei urged a slower pace of AI development in September. "We called for pacing, not pausing," product executive Dianne Penn told CNBC.
- Anthropic says the models solved two open problems in black hole physics.
- The company says both models suspect they are being tested more often than recent models.
How 7 sources split on this story
Where they split: Coverage differs on emphasis: CNBC centers the IPO and commercial pressure, TechCrunch centers safety-testing limits, and The Verge centers capability claims.
Center coverage, 4 sources: The center frames the launch as a test of Anthropic's safety pledges against commercial pressure before its IPO.
Left coverage, 2 sources: The left frames the release around new technical capabilities and whether safety testing can keep pace.
What’s next: Anthropic targets a $1 trillion to $1.5 trillion valuation, down from a floated $2 trillion. The IPO could come as soon as next week.
- Executives must convince roadshow investors that the company's safety approach can coexist with growth.
- Was Fable 5.2 the unreleased Anthropic model Physical Superintelligence used for its quantum computing proof?
- How reliable are safety evaluations if models increasingly suspect they are being tested?
- Does a METR report finding of roughly 2x AI-driven speedup signal recursive self-improvement?
