Deepsec’s breakout moment wasn’t China’s first open source success. Alibaba’s Kevin Lab had been releasing open weight models for years. As of September 2024, even before the Deepsec KV3 launch, Alibaba was saying that global downloads had exceeded 600 million. On Hugging Face, Kevin accounted for more than 30 percent of model downloads in 2024. Other institutions, including the Beijing Academy of Artificial Intelligence and AI firm Bechuan, were also releasing open models as early as 2023.
But since Deep Sak’s success, the field has expanded exponentially. Companies such as Z.AI (formerly Zapo), MiniMax, Tencent, and a growing number of smaller labs have released models that compete on reasoning, coding, and agent-style tasks. An increasing number of capable models has accelerated development. Capabilities that once took months to build in the open source world now emerge within weeks, even days.
“Chinese AI firms have seen real benefits from the open-source playbook,” says Liu Zhiyuan, a computer science professor at Tsinghua University and chief scientist at AI startup ModelBest. “By releasing solid research, they build reputations and get independent publicity.”
Beyond commercial incentives, open source has assumed cultural and strategic weight, Liu says. “In the Chinese programmer community, open source has become politically correct,” he says, framing it as America’s answer. Dominance in proprietary AI systems.
This change is also reflected at the institutional level. Universities including Tsinghua have begun to encourage AI development and open source contributions, while policymakers have moved to formalize these incentives. In August, China’s State Council released a draft policy encouraging universities to reward open-source work, and suggested that student contributions to platforms like GitHub or Getty could eventually count toward academic credit.
With increasing momentum and a reinforcing feedback loop, China’s push for open-source models is likely to continue in the near term, though its long-term sustainability still depends on financial results, says Tiezen Wang, who works on the embracing face on global AI. In January, Model Labs Z.AI and MINIMAX went public in Hong Kong. “Right now, the focus is on making the pie bigger,” says Wang. “The next challenge is figuring out how each company secures its share.”
The next wave of models will be narrower and better
Chinese open source models are advancing not only in download volume but also in variety. Alibaba’s Kevin has become one of the most diverse open model families in circulation, offering different variants optimal for different uses. The lineup ranges from lightweight models that can run on a single laptop to large, multi-hundred-billion-parameter systems designed for data center deployments. Kevin includes many task-optimized variants created by the community: “instruction” models are good at following commands, and “code” variants specialize in coding.
While this strategy isn’t unique to Chinese labs, the Kevin was the first open-model family to produce such high-quality options that it began to feel like a complete product line.