Report: Chinese Models Account for 89% of Global Open-Source Model Enterprise Spending
Menlo Ventures reports that Chinese models account for 89% of global enterprise spending on open-source models. Kimi leads with a 61% share, followed by DeepSeek at 15%, Z.ai's GLM series at 11%, and Meta's Llama at 7%. Open-source models represent only 4% of total model spending, while Anthropic and OpenAI together command 86% via API. The data comes from a September survey of nearly 400 U.S. business decision-makers.
A report released this week by Menlo Ventures states that Chinese models account for 89% of global enterprise spending on open-source models. Moonshot AI's Kimi leads with a 61% share, followed by DeepSeek at 15%, Z.ai's GLM series at 11%, and Meta's Llama at just 7%.
However, open-source models represent only 4% of total model spending, with Anthropic and OpenAI collectively capturing 86% through their APIs. This data comes from a September survey of nearly 400 business decision-makers in the United States.
South China Morning Post