Over the past 18 months, two generations of China’s open-source AI models have put sustained pressure on the United States.
一年半的时间，中国连续两款开源AI让美国感到了压力。

The first wave came in January 2025. DeepSeek overturned the industry consensus that advanced AI requires massive capital investment and endless computing spending.
第一轮，是2025年1月份，DeepSeek颠覆了“AI等于烧钱”的行业叙事。

The second milestone arrived on July 17. Kimi K3 became the first open-source model to outperform leading closed-source models in front-end programming benchmarks.
第二轮，7月17日，Kimi K3诞生，首次有开源模型在前端编程领域超越一众闭源模型。

Unlike Silicon Valley’s closed proprietary strategy, these Chinese models adopt open-source frameworks, enabling global developers to deploy and adjust models locally.
与硅谷闭源路线不同，这两款中国模型选择开源路线，允许全球开发者本地部署、自主微调。

The two successive breakthroughs prove that China’s open-source AI ecosystem is delivering steady, iterative progress rather than occasional surprises.
两轮接续突破证明，中国开源AI生态正在持续迭代，不再只是偶然的技术惊喜。

核心词汇

1. open-source model 开源模型
​
2. closed-source / proprietary model 闭源模型
​
3. computing spending 算力投入
​
4. front-end programming 前端编程
​
5. benchmark 评测基准
​
6. iterative progress 持续迭代
ai发展路线 对ai的介绍 AI技术解读