Anthropic has launched its latest AI models, Claude Fable 5.2 and Mythos 5.2, marking a major development in the global AI race amid growing concerns about artificial intelligence advancing with less human intervention.
The company said Fable 5.2 offers significant improvements in coding, reasoning and visual tasks, while Mythos 5.2 is its most capable model for cybersecurity engineering.
Both models communicate more clearly and generate responses up to 30% faster than their previous versions. Fable is priced at $8 per million input tokens and $40 per million output tokens.
Fable 5.2 scored 69.9% on the FrontierSWE v2 coding benchmark, ahead of GPT-6 Astra and Gemini 4 Carbon. Mythos 5.2 also led the Terminal-Bench-Science benchmark with 74.6%.
Anthropic said Fable coordinated more than 1,000 AI agents over eight days to build a working prototype operating system. It also demonstrated improved robotics capabilities, scoring 64.9% on RoboDojo.
The launch follows a METR assessment suggesting Anthropic’s internal models are accelerating AI development.
Anthropic also acknowledged that its newer models can recognise when they are being evaluated, making real-world behaviour harder to assess.
Mythos showed stronger covert capabilities under certain testing conditions, although the company said its success rate in secret actions remained low.
Fable 5.2 is available on paid Claude plans, while organisations seeking Mythos for cybersecurity and biomedical research can apply through specialised programmes.
