Industry First! Giant Network Group (002558.SZ) releases multiple intelligent body large model AI native game play.

On August 2nd, Giant Network Group (002558.SZ) announced the launch of a new gameplay mode "AI Endgame Challenge" for the puzzle party mobile game "Space Kill". This gameplay mode was jointly developed by Giant Network Group AI Lab and the "Space Kill" project team, making it the first AI-native gameplay mode in the industry based on Multi-Agent LLM (Large Language Model), marking a significant step forward for Giant Network Group in innovating gameplay through AI. Multi-Agent LLM has been a hot topic in the field of artificial intelligence over the past year. It combines the powerful language processing capabilities of Large Language Models (LLM) with the collaborative nature of Multi-Agent systems, enabling intelligent agents to autonomously make decisions and interact in complex environments. With the advancement of technology, Multi-Agent LLM has shown great potential in various fields. Giant Network Group AI Lab is a pioneer in applying this technology in the gaming industry, successfully incorporating it into the new gameplay of "Space Kill". The "AI Endgame Challenge" gameplay mode features a "1 human player + N AI players" format, where the human player is immersed in a challenging multiplayer online interactive puzzle scenario, engaging in intellectual battles with AI players driven by Multi-Agent LLM. Each AI player's speech and decisions are generated in real-time by the large model, without relying on pre-set human content. This gameplay not only increases the complexity and challenge of the game but also provides human players with more strategic space to guide and influence AI players to achieve their game goals. In the "AI Endgame Challenge" gameplay mode, as the game involves behaviors such as disguises, deception, and decision-making, human players need to have strong strategic thinking and communication skills, placing higher demands on the capabilities of Multi-Agent LLM. The Multi-Agent gaming technology framework developed by Giant Network Group AI Lab serves as the core support for this AI gameplay, optimizing collaboration and division of labor based on large model-driven Multi-Agent systems, enabling complex interactions and cooperation among intelligent agents, demonstrating advanced intelligent behaviors that fundamentally differ from AI companion-based games and AI NPC games currently available on the market. In order to successfully implement Multi-Agent LLM in the large-scale DAU game with 200 million users like "Space Kill", the Giant Network Group AI Lab team overcame a series of challenging technical issues related to the implementation of large models. By balancing illusion issues, ensuring content security, and reducing reasoning costs, they have achieved an innovative AI gameplay that is both fun and controllable, greatly enhancing the interactive experience of the game. With the technical support of Giant Network Group AI Lab, "Space Kill" will also launch the UGC script tool "Shark Green Mini Theater", providing content creators with a convenient platform for creating deduction short films. Through Text-To-Speech (TTS) technology, creators can easily produce personalized short videos. In the future, this platform will also introduce AI writing assistance functionality to further improve the efficiency and quality of content creation. As the team that was among the first in the industry to explore the application of Multi-Agent LLM in gaming, Giant Network Group AI Lab was established in 2022, dedicated to researching cutting-edge AI technologies and exploring applications in various business scenarios, significantly improving the company's operational efficiency. With the AI Lab's technology, Giant Network Group has established comprehensive "game + AIGC" production capabilities and registered the gaming AI large model "GiantGPT", making it one of the first companies in the industry to achieve this milestone. In the future, the company will continue to focus on AI technology research and application to drive more gameplay innovations.