Anthropic unveils claude fable 5.2, outperforming gpt-6 astra in coding and benchmarks

Anthropic announced the release of two advanced AI models, Claude Fable 5.2 and Claude Mythos 5.2, on November 10, 2026. The company claims these new models show significant improvements across coding, reasoning, scientific research, cybersecurity, and robotics, while also boosting output speed by 30% and reducing operational costs by 20%. Performance metrics highlight Fable 5.2's strength. In the FrontierCode 2.0 benchmark for agent coding, it achieved 49.1%, surpassing OpenAI's GPT-6 Astra (42.8%), Google's Gemini 4 Carbon (41.1%), and Anthropic's Opus 5.5 (45.5%). Furthermore, Fable 5.2 scored 74.2% on the Humanity's Last Exam, leading GPT-6 Astra's 57.2%. Other notable capabilities include Fable 5.2 coordinating over 1,000 AI agents to build a bootable operating system over eight days. In the VIALS-AA scientific visual reasoning test, Fable 5.2 scored 67%, higher than Gemini 4 Carbon (64%) and GPT-6 Astra (63%). On the AA-3D benchmark for interactive 3D experience building, Fable 5.2 (Max) achieved an Elo score of 1,271, significantly higher than Claude Sonnet 5.5 (1,204) and Claude Opus 5.5 (1,202).