Agentic data science with the Data Agent Kit
cloud.google.com·6h ago
Data Science
Interconnects.ai·21d ago·5 min read
The current balance of power in open models
TL;DR
Open-weight models, primarily led by Chinese companies, dominate the landscape of AI, contrasting with U.S.-developed open-source models that include full reproducibility details. This shift highlights the competitive dynamics in AI development between the U.S. and China.
✦ Why It Matters
Evaluate your AI strategy to incorporate open-source models for better reproducibility and innovation.
Key Takeaways
- Monitor the performance of Chinese open-weight models in your projects.
- Consider partnerships with Chinese model labs for competitive advantages.
- Invest in open-weight models to enhance your AI capabilities.
- Stay informed about regulatory changes affecting open models.
Full Summary
Open-weight models, which allow public access to their underlying weights and code, have seen a dramatic rise in usage, particularly from Chinese companies like Alibaba and Z.ai. As of September 2026, Chinese models dominate the market, accounting for over 80% of usage on platforms like OpenRouter, with significant adoption in academia and industry.
For instance, mentions of Chinese models in AI research papers have surged to over 40%, surpassing American models at 30%. The competition is intensifying, with Chinese models like GLM-5.3 and Kimi K3 outperforming American models in key benchmarks.
Despite American efforts to catch up, the gap in capabilities between open and closed models is narrowing, with Chinese models being released faster and focusing on high-demand tasks. This shift raises concerns about cybersecurity risks and regulatory challenges, necessitating a proactive approach to managing open-weight models.