Anthropic releases Claude Fable 5.2 and Claude Mythos 5.2, cuts prices and improves output speed
According to an announcement on Anthropic’s official website, Anthropic has released Claude Fable 5.2 and Claude Mythos 5.2. The two are the same underlying model. Fable comes with cybersecurity and biological safeguards and is open to all users; Mythos must be applied for through a cybersecurity or life sciences verification program.
The new models have improved in programming, reasoning and vision, among other areas, and output speed is up by as much as 30% compared with version 5.1. API prices have been reduced to $8 per million input tokens and $40 per million output tokens (previously $10 and $50 for Fable 5.1). The company said Fable 5.2 scored 49.1% on the FrontierCode 2.0 benchmark (Xhigh effort), the highest among the released models it has tested. Fable 5.2 is now available on the Claude platform and on AWS, Google Cloud and Microsoft Azure.
Original source:
Title: Introducing Claude Fable 5.2 and Claude Mythos 5.2 Publisher: Anthropic Publication date: 2026-11-09 URL: https://www.anthropic.com/claude-fable-and-mythos-5-2 Source type: official company announcement Verification status: the full text was actually read in this instance, verification date 2026-11-09, no access restrictions Supporting content: supports that the two are the same underlying model, that Fable is open to all users and equipped with cybersecurity and biological safeguards, that Mythos requires application through a verification program, that output speed is up by as much as 30%, that prices were reduced to $8/$40, that FrontierCode 2.0 (Xhigh) scored 49.1% as the highest among released models, and that it is available on the Claude platform and on AWS, Google Cloud and Microsoft Azure
ABAB AI Insight
Anthropic split the same underlying model into two versions for release: Fable 5.2 retains cybersecurity and biological safeguards and is aimed at all users; Mythos 5.2 reduces safeguards in those areas and is open only to institutions that pass the Cyber Verification Program or Life Sciences Verification Program. This design continues the company’s earlier approach in the Fable/Mythos line, separating specific-domain risks from general availability for a high-capability model.
On pricing, the input and output token prices both fell 20% from Fable 5.1 (from $10/$50 to $8/$40). The announcement also provides an optional “fast mode” that can increase token generation speed by up to 2.5 times, priced at $16 per million input tokens and $80 per million output tokens. The announcement says that in the FrontierCode 2.0 benchmark, Fable 5.2 scored 49.1% under Xhigh effort, higher than its own 5.1 version and compared models Opus 5.5, GPT-6 Astra and Gemini 4 Carbon.
All statements about performance improvements, speed and benchmark scores above are from Anthropic itself. The announcement also mentions pre-release tests in which the model completed software migration, designed electronic devices, solved open problems in black hole physics and proved Fermat’s Last Theorem formally, but it did not provide independently reproducible detailed data. Observable follow-up developments include whether third-party evaluations reproduce the FrontierCode 2.0 scores and the actual application and approval status of the Mythos 5.2 verification program.
ABAB News · Cognitive Law
Fable and Mythos are the same underlying model, with the main difference being whether cybersecurity and biological safeguards are enabled.
The 20% price cut and speed increase are company-reported, and the 49.1% FrontierCode 2.0 score also comes from Anthropic’s own testing.