Cai Haoyu's AI company Anuttacon releases its first video large model LPM1.0

ME News Report, April 11 (UTC+8), the AI company Anuttacon founded by Mihoyo’s founder Cai Haoyu quietly released a video character performance large model LPM 1.0, supporting real-time full-duplex audio and video conversations, capable of simultaneously speaking, singing, listening, and emotional reactions, with lip-sync, micro-expressions, and body rhythm details reaching a high level of realism. Technically, LPM 1.0 uses a 17B parameter base model distilled into a low-latency streaming online version, supporting arbitrary-length identity-consistent generation, compatible with realistic, 2D anime, 3D game, and non-humanoid character styles, without fine-tuning. The team also launched the character performance evaluation benchmark LPM-Bench, achieving SOTA results on multiple metrics. Compared to Kling-Avatar 2.0 and OmniHuman 1.5, which have a maximum 30-second generation limit, LPM 1.0 supports unlimited duration output, targeting dialogue agents, game NPCs, and virtual live streaming scenarios. Notably, Anuttacon explicitly states that this release is solely for academic exchange, with no plans to open source or provide APIs, demos, or any products externally. The general consensus is that one of the main purposes of this public release is to attract AI talent. LPM 1.0 is Anuttacon’s latest move in the AI game engine and intelligent NPC track since its founding, and also another milestone in Cai Haoyu’s gradual deepening of the underlying model capabilities from Mihoyo’s AI tool development to the core model of AnuNeko chat products. (Source: BlockBeats)