The artificial intelligence race may no longer be defined by the most powerful frontier models. New data shows that open-weight models — particularly those from Chinese developers — are capturing an increasing share of production workloads, challenging the dominance of proprietary systems from US labs such as OpenAI and Anthropic.
According to figures from Hugging Face, a leading platform for hosting and deploying open models, Chinese open-weight models accounted for 41% of all downloads on the platform this spring, overtaking US models for the first time. On the routing platform OpenRouter, six of the top seven most popular models come from Chinese firms including Tencent, Xiaomi, DeepSeek, MiniMax, and Z.ai. Anthropic's Claude Opus 4.7 sits in seventh place. Separately, data from Vercel indicates that open-weight models handled nearly a third of all AI requests on its platform in June, absorbing the high-volume infrastructure layer while closed models operate as a premium tier.
Hugging Face CEO Clem Delangue argues that the trend reflects a fundamental shift in enterprise strategy. 'If you're an AI company or a technology company, you don't want to outsource your core capabilities to another company, to a black box API that you don't control,' he said. Delangue noted that a new repository is created every seven seconds on the platform, which now hosts almost three million public models and one million public datasets. Half of all Fortune 500 firms use Hugging Face to deploy private or open-source models, he added.
The shift has significant implications for UK businesses. Open models offer lower deployment costs and greater flexibility, allowing firms to customise AI for specific use cases without the recurring fees associated with proprietary APIs. However, they also raise questions about data sovereignty and regulatory compliance. The UK Information Commissioner's Office (ICO) has not yet issued specific guidance on open-weight models, though the EU AI Act imposes transparency obligations on providers of general-purpose AI, which could cover some open-source releases. Microsoft CEO Satya Nadella has also warned against single-provider lock-in, urging enterprises to retain control of their data.
For UK consumers, the growing availability of open models could mean cheaper AI-powered services and faster innovation in sectors such as healthcare, finance, and retail. However, experts caution that open models may carry higher risks of misuse or bias if not properly governed. Delangue predicts that frontier models will increasingly be reserved for specialised, high-value tasks, while the majority of production workloads will run on open or private models. 'Maybe in a few years, the frontier models will be for experimenting and some really high-value tasks,' he said.
The trend is accelerating as Chinese labs release increasingly capable open-weight models. Most recently, Beijing-based Z.ai launched GLM-5.2, which competes with Anthropic's latest models on agentic coding and security vulnerability detection. For UK policymakers, the shift underscores the need for a balanced regulatory approach that fosters innovation while addressing risks around transparency, safety, and competition.