LPM 1.0

LPM 1.0 (Large Performance Model) represents a revolutionary advancement in real-time AI video character performance generation. Developed by Anuttacon, this 17B-parameter Diffusion Transformer creates conversational agents, game NPCs, and streaming characters that speak, listen, sing, and react with human-like nuance. Unlike previous robotic avatar systems with looping animations and frozen expressions, LPM 1.0 achieves full-duplex conversation—characters generate video while simultaneously listening and reacting to user input. The model delivers identity-consistent performances lasting over 45 minutes without drift, supports zero-shot generalization across photorealistic humans, anime, 3D characters, and non-humanoid styles, and operates with just 0.35-second latency—three times faster than alternatives. Its multi-granularity identity conditioning system preserves character details without hallucination, while multimodal control through text, audio, and image references enables fine-grained artistic direction. LPM 1.0 serves as a visual engine for next-generation interactive experiences across conversational AI, gaming, live streaming, education, and accessibility applications.