AI 导读
视频模型能生成，但要让你等。 实时虚拟形象能响应，但有局限。 游戏引擎能实时运行，但依赖预制内容。 R2 指向第四种范式：为持续交互而构建的可扩展生成式世界。https://t.co/OyCAFIAfXP
正文
Video models generate, but make you wait.
Real-time avatars respond, but within limits.
Game engines run live, but rely on prebuilt content.
R2 points to a fourth paradigm: scalable generative worlds built for continuous interaction. https://t.co/OyCAFIAfXP
来源：@PixVerse_ · x.com