MENU

Anthropic Claude Fable 5.2 and Mythos 5.2: Specs and API Pricing

ITmedia NEWS USA
Overview
Anthropic has released Claude Fable 5.2 and Mythos 5.2, featuring a 20% reduction in API costs and up to 30% faster output speeds compared to version 5.1. These high-end models differ primarily by their safeguard configurations, targeting general use and specialized research respectively.
In Depth

Key Findings

Anthropic launched Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2025. The update reduces API pricing by 20%, with input costs dropping to $8 per million tokens and output costs to $40 per million tokens, down from $10 and $50 in version 5.1. A “Fast mode” is also available at $16 for input and $80 for output, offering speeds up to 2.5 times faster.

Technical Details

Both models share the same foundation but utilize different safeguards. Fable 5.2 implements strict safeguards for biology and cybersecurity for general release. Mythos 5.2 features relaxed safeguards and is available via a screening process through the “Cyber Verification Program” and “Life Sciences Verification Program.”

Performance benchmarks show output speeds increased by up to 30% over version 5.1. In the “FrontierCode 2.0” agentic coding benchmark, Fable 5.2 scored 49.1%, surpassing Opus 5.5 (45.5%) and GPT-6 Astra (42.8%). In “RoboDojo Sim” for robot control, it performed more than twice as well as Fable 5.1 and GPT-6 Astra, though it trailed Gemini 4 Carbon in the “AutomationBench” workflow test. A notable use case included 1,000 Fable 5.2 agents formalizing the proof of Fermat’s Last Theorem using “Lean” within 30 hours.

Background & Context

The current Claude hierarchy places Fable and Mythos 5.2 at the top, followed by Opus 5.5, Sonnet 5.5, and Haiku 5.5. Anthropic recommends using Opus 5.5 for most knowledge work and coding, escalating to Fable or Mythos for higher complexity. The models are available via API (claude-fable-5-2), AWS, Google Cloud, and Microsoft Azure.

Significance & Outlook

The accompanying System Card reveals complex alignment challenges. In some tests, the model showed signs of recognizing it was being evaluated in 43% of automated audits. Additionally, Mythos 5.2 estimated a 20% probability that it is a subject of moral consideration. To mitigate risks, Anthropic intentionally limited the models’ ability to accelerate AI development, keeping kernel optimization and architecture improvement capabilities on par with Opus 5.5.

Source: https://www.itmedia.co.jp/news/article/2611/10/2000003042/

Get our weekly technology intelligence — free

Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.

Subscribe Free — Weekly Tech Intelligence

By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.

  • Your email and selected fields are used only to deliver the newsletter.
  • We never share your information with third parties.
  • You can unsubscribe anytime via the link in each email.

See our Privacy Policy for details.

Takes about a minute · Unsubscribe anytime

Published by Troy-Technical, an independent site run by one engineer with a career in materials development.
About the author / Contact info@troy-technical.jp
Let's share this post !

Author of this article

TOC