The Rise of Open AI Models: Are Frontier Models Losing Their Edge?
Open-weight AI models are surging in popularity, outpacing closed competitors, as developers and enterprises favor customizable and cost-effective alternatives to proprietary systems.
While the AI industry focused on Anthropic's latest frontier models and Washington's access restrictions, developers quietly adopted open models at a rapid pace. Chinese open-weight models accounted for 41% of downloads on Hugging Face this spring, surpassing US models. On OpenRouter, six of the top seven models are open models from Chinese firms like Tencent, Xiaomi, DeepSeek, MiniMax, and Z.ai. Anthropic's Claude Opus 4.7 sits in seventh place.
Data from Vercel shows open models handled nearly a third of AI requests on the platform in June, acting as a high-volume, low-cost infrastructure layer, while closed models serve as a premium tier. Although these platforms don't capture all AI usage, the growing market share of open models raises questions about the continued relevance of frontier models for most production workloads.
Clem Delangue, CEO of Hugging Face, notes that enterprises increasingly prefer owning their AI models rather than renting them. He points to Hugging Face’s activity: a new repository is created every seven seconds, the platform hosts nearly 3 million public models, and half of Fortune 500 companies use it to deploy private or open models.
Delangue argues that the biggest risk in AI is concentration of power. He advocates for transparency and leveling the playing field, stating that keeping models closed does not make them safer but rather creates an asymmetry of power and capabilities.
Microsoft CEO Satya Nadella also warns against single-provider lock-in, emphasizing the importance of data control for enterprises. Meanwhile, Chinese AI labs continue to release powerful open models. For instance, Z.ai recently launched GLM-5.2, which excels at agentic coding and matches Anthropic's latest models in identifying security vulnerabilities.
