Rednote joins wave of Chinese firms releasing open-source AI models

In a significant move within the tech landscape, Rednote, one of China's leading social media platforms, has launched an open-source large language model. This initiative positions Rednote among a growing number of Chinese technology companies that are opting to make their artificial intelligence innovations publicly accessible. This strategy stands in stark contrast to many American tech giants, such as OpenAI and Google, which have chosen to keep their most advanced AI models proprietary. However, some U.S. companies, including Meta, have embraced open-source models as well. By open-sourcing their AI technology, Chinese firms can showcase their capabilities, cultivate developer communities, and extend their global influence, particularly as the U.S. imposes restrictions on the export of advanced semiconductor technologies. Rednote's newly introduced model, named dots.llm1, is now available for download on the developer platform Hugging Face. Accompanying this release is a technical paper that elaborates on the model's specifications. In practical applications, the model has shown competitive performance in coding tasks, comparable to Alibaba's Qwen 2.5 series, although it falls short in comparison to more advanced models like DeepSeek-V3, according to the technical documentation. Often referred to by its Chinese name, Xiaohongshu, Rednote is akin to Instagram, allowing users to share a variety of content including photos, videos, and live streams. Earlier this year, the platform gained international traction as U.S. users began flocking to it amid uncertainties surrounding a potential TikTok ban. Since 2023, Rednote has been heavily investing in the development of large language models, a venture that gained momentum following the release of OpenAI's ChatGPT in late 2022. Recently, the company has ramped up its AI initiatives by launching Diandian, an AI-driven search application designed to help users navigate content on its primary platform. Other notable players in the open-source AI space include Alibaba, which introduced an upgraded version of its model, Qwen 3, in April, and the startup DeepSeek, which earlier this year released its low-cost R1 model as open-source software, disrupting the global AI market with its effective performance at a fraction of the cost of Western counterparts.