Chinese social media giant Rednote, widely known by its Chinese name Xiaohongshu, has released its first open-source large language model.
It joins a growing list of Chinese tech firms making strides in artificial intelligence by offering their models to the global developer community.
The model, named dots.llm1, is now available for public download on the AI developer platform Hugging Face, with a technical paper also published to provide performance benchmarks and design insights. This move reflects China’s broader strategy to showcase its AI capabilities, build influence, and expand developer ecosystems despite U.S.-led restrictions on advanced chip exports.
Unlike many U.S. tech giants such as OpenAI and Google, which have largely kept their most sophisticated AI models proprietary, Rednote’s open-access strategy follows in the footsteps of other firms like Meta, Alibaba, and DeepSeek, who have embraced open-source releases in a bid to democratize access and boost innovation.
According to Rednote’s technical documentation, dots.llm1 performs comparably to Alibaba’s Qwen 2.5 series in coding-related tasks, although it lags behind more advanced Chinese models such as DeepSeek-V3. Nevertheless, the launch marks a significant milestone for the social media firm, which only began investing in large language models in 2023, shortly after the global rise of OpenAI’s ChatGPT.
Rednote, often dubbed “China’s Instagram,” is a lifestyle and content-sharing platform that enables users to post photos, videos, text updates, and livestreams. The app recently gained attention from international audiences, particularly in the United States, where some users began migrating to Rednote amid rising concerns over a potential ban on TikTok.
In recent months, Rednote has ramped up its AI development efforts, notably launching Diandian, an AI-powered search tool designed to enhance content discovery on its main platform.
The company now joins a broader trend of Chinese firms pushing boundaries in the open-source AI space. In April, Alibaba released Qwen 3, the latest version of its foundational model, while DeepSeek, a startup, shook the industry with its high-performing yet cost-effective R1 model.
As global competition intensifies in the AI sector, China’s open-source approach is increasingly being viewed as a strategic move to sidestep technological isolation and foster a more collaborative innovation environment—both at home and abroad.
