大模型是人工智能技术的最新成就，开源创新逐渐成为大模型发展的新引擎。人工智能开源实践的发展，冲击着西方学术传统中商业公司与开源社区合作达成公地治理的经典开源范式。在 20 世纪七八十年代个人计算机兴起的过程中，开源尚未获得商业成功，而在 21 世纪初开启的互联网时代，开源与商业结合并发展出全球化的互联网产业。在演进历程中，开源创新的四大价值逻辑发挥了关键作用：开源是应对技术发展不确定性的重要创新模式，是吸引各类型创新主体参与贡献并激发创新活力的重要生态途径，是破除行业垄断的有效商业策略，是保障新兴产业安全发展的重要依托。大模型开源治理继承了开源创新的价值逻辑，同时也面临挑战，原因在于开源大模型与开源软件相比，开源内容有较大不同，开源治理维度更复杂，其背后引发的人工智能技术竞争格局的调整更为深刻。而以深度求索系列开源大模型为新起点，中国更有可能通过建立有为政府、有效市场和有机社会的协调机制，构建新型的大模型开源创新公地。

Large models represent the cutting-edge advancements in artificial intelligence technology, with open source innovation increasingly emerging as a critical driver of their development. The evolution of artificial intelligence open-source practices challenges the classical paradigm of public governance within the Western academic traditions, which emphasize collaboration between commercial entities and open-source communities. The intrinsic four major value logics of open source—focusing on technological innovation, ecological vitality, business strategy, and security assurance—have played a pivotal role in its transition from “struggle” to “prosperity”. At the same time, the open-source governance of large models is also facing challenges. This is because, compared with open-source software, the open-source content of large models is quite different, the dimensions of open-source governance are more complex, and the adjustment of the competitive pattern of artificial intelligence technology triggered behind it is more profound. Taking the DeepSeek series of open-source large models as a starting point, China is more likely to build a new model of open-source innovation commons for large models by establishing a coordination mechanism among a proactive government, an effective market, and an organic society.

artificial intelligence, open-source governance, large models, open-source innovation, commons, Chinese narrative