Giant Network releases the first DeepSeek native gameplay "Insider Challenge"

On March 14, Giant Network announced that its social inference game "Space Kill" was officially connected to the DeepSeek big model and launched the native gameplay "Insider Challenge" based on this technology. The gameplay has now started grayscale testing and will be open to all users in the future. This marks the first time in the industry that DeepSeek's big model capabilities have been deeply applied to gameplay innovation, using AI to drive the core gameplay and reshape the game's competitive and interactive experience.

The gameplay of "Insider Challenge" is based on the core reasoning mechanism of "Space Kill". Each game is fought against two real players and eight AI players powered by DeepSeek. With its powerful reasoning capabilities, low latency and vertical optimization technology, DeepSeek can drive eight AI players to compete fiercely with real-life players at the same time. In the highest difficulty mode, AI players show superb logical reasoning skills, can accurately analyze loopholes in conversations, and even eliminate real-life players through alliances.

Unlike the existing DeepSeek game assistant, Space Kill reconstructs the game's interactive experience and achieves innovative gameplay by introducing new AI players powered by DeepSeek. The intelligent system built by AI players based on DeepSeek's strong logical reasoning and realistic interaction capabilities has an essential difference from tool-type AI assistants. Its human-like logical offensive and defensive capabilities, the dynamic competitive relationship formed with real-life players, and the millisecond real-time interaction significantly improve the authenticity of the confrontation. With the help of the reinforced learning framework, AI players can generate intelligent strategy game mechanisms such as disguise, accusation, and countermeasures, which not only enhances the challenge of the game, but also optimizes the reasoning interaction experience between players and redefines the intelligent boundaries of human-computer confrontation.