Anthropic's new generation model, Claude Fable 5.5, appears to have been released to a select group of users. AI researcher Chetaslua demonstrated the model on November 2nd (Beijing time), stating that even with internet search and memory disabled, the new model could still answer questions about obscure events in 2026. When generating animations using the same prompts, its performance outperformed the regular Opus 5.5 and Fable 5.1. More notably, Fable 5.5 autonomously worked continuously for approximately 2 hours and 40 minutes, completing a 66-second animated short film using only Python code. This film included characters, scenes, frame-by-frame animation, subtitles, music, and sound effects, ultimately outputting a single 1920×804, 24 frames per second HTML file. The author stated that the entire process did not utilize image generation models, image libraries, audio samples, or external fonts. This suggests that the new model's capabilities may extend to long-duration autonomous execution, complex visual content creation, and audio-visual co-generation, beginning to demonstrate the ability to independently complete complete creative projects.
