Anthropic Launches Claude 5.2 Fable and Mythos Models with Advanced Coding and Scientific Capabilities

According to Anthropic News:

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, sharing the same underlying architecture tailored for the most demanding technical workloads. During testing, the models helped engineers perform complex software migrations, design electronic devices, and accelerate physics research. Benchmark results show Fable 5.2 outperforming previous iterations and competitors in agentic coding, hardware engineering, and multidisciplinary reasoning. Notably, an engineer used Fable 5.2 to coordinate over 1,000 subagents to build a working prototype operating system in eight days.

In science and mathematics, Fable 5.2 resolved open theoretical physics problems related to black holes and formalized the proof of Fermat’s Last Theorem in 30 hours. Meanwhile, Mythos 5.2 serves as the company's most capable cybersecurity model, helping companies like Mozilla and Rubrik discover hidden software vulnerabilities more effectively. Despite these advanced capabilities, Anthropic reports that safety and alignment risks remain low, backed by extensive pre-release testing and improved efficiency features such as an optional fast mode.