Anthropic's Claude Fable 5.2 Beats GPT-6 Cyber in Multi-Stage Cyberattack Benchmark

Anthropic has released updated models, Claude Fable 5.2 and Claude Mythos 5.2, which show improvements in areas such as coding, reasoning, and vision. These large language models were used during their training to assist engineers with various tasks. The models helped complete complex software migrations and design electronic devices from scratch. They also contributed to solving open problems in physics related to black holes. Additionally, the models were tested in a multi-stage cyberattack benchmark where Claude Fable 5.2 outperformed GPT-6 Cyber. The updated models demonstrate advanced capabilities in multiple areas.