Krafton unveils co-playable character powered by Nvidia Ace

Co-playable characters in Krafton’s upcoming life simulation sport InZOI / Captured from Nvidia’s YouTube

Krafton unveiled an agentic framework in collaboration with Nvidia at CES 2025 in Las Vegas, Tuesday (native time).

The partnership combines Krafton’s famend sport growth experience with Nvidia’s cutting-edge AI expertise. Krafton showcased an AI innovation set to redefine the way forward for the worldwide gaming business.

A co-playable character (CPC) is an AI companion designed to work together with avid gamers, powered by an on-device small language mannequin for gaming constructed with Nvidia Ace expertise. In contrast to conventional non-player characters, CPCs can have interaction and cooperate with customers, understanding and responding to sport conditions in a human-like method.

Krafton plans to increase the usage of CPC expertise throughout its gaming portfolio, together with the PUBG IP franchise and InZOI, to reinforce the consumer expertise.

“CES 2025 marks a big milestone for Krafton as we work to unveil AI expertise that may revolutionize the gaming business,” Krafton CEO Kim Chang-han mentioned.

“We plan to intently collaborate with Nvidia to drive the transformation of the consumer expertise with AI-powered innovation constructed with Nvidia Ace, akin to CPC, that’s poised to redefine the way forward for gaming.”

Keita Iida, vp of developer relations at Nvidia, mentioned, “AI is reworking the best way video games are created and performed. By bringing our progressive AI expertise, akin to Nvidia Ace, to a variety of their reside video games, Krafton is unlocking co-playable characters for brand new, distinctive experiences.”

Since establishing its Deep Studying Division in 2022, Krafton has secured key AI applied sciences, together with pure language processing, 3D imaginative and prescient and animation, voice synthesis (STT/TTS), reinforcement studying and multi-modal studying. The corporate additionally showcased its R&D capabilities by presenting a number of papers at main AI conferences, together with NeurIPS, ICML and ICLR.