GitCode launches DeepSeek full package! Leading the new journey of domestic large models

This Spring Festival, the AI ​​field is very lively. DeepSeek has been "soaring" all the way, causing huge waves in the US technology community and the US stock market . On January 27, the three major US stock indexes plummeted at the opening. Major US technology stocks such as Nvidia, Microsoft, Google's parent company Alphabet, and Meta all suffered stock market earthquakes. Among them, Nvidia fell nearly 17%, and its market value evaporated by about US$600 billion in a single day, setting a record for US stocks.

With its open source features and extremely low cost, DeepSeek has performed well in tasks such as mathematics, programming, and natural language reasoning, and its performance is no less than that of the top AI models in the United States. In particular, DeepSeek-R1 has achieved a leap-forward improvement in reasoning capabilities with very little labeled data through the innovative use of reinforcement learning technology. In core capabilities such as mathematics, programming, and language understanding, it is perfectly comparable to OpenAI o1 . This series of innovative achievements has not only been widely recognized by the industry, but also allowed the world to see the rise of China's AI technology.

In recent years, China's AI has developed rapidly, with full blossoming in basic research, technological innovation, and application implementation. More and more domestic large models are emerging. As a leading code hosting open source community, GitCode has been growing together with China's AI industry, providing the ultimate community experience and developer tools for developers around the world.

We are well aware of developers' desire for powerful and convenient development tools, and we are also actively involved in the construction of the domestic large model ecosystem. The launch of the domestic large model channel and the introduction of the DeepSeek family bucket are an important step for GitCode to fulfill its mission of promoting the development of China's AI. It aims to provide developers with a better development environment and help China's AI industry reach new heights.

The GitCode homepage has launched a new section dedicated to domestic models. It is a gathering place for large domestic models, opening the door for developers to easily obtain various high-quality models.

GitCode's domestic large model channel integrates rich model resources including the DeepSeek family bucket, and provides a local model download function. Developers do not need to rely on network calls all the time. Not only can they use the model normally in an offline environment, avoiding the impact of network fluctuations on development, but also reducing data transmission risks and ensuring data security and privacy. At the same time, the channel brings together domestic large models in multiple fields. Users can freely choose according to their own project characteristics and needs, avoiding the limitations of a single model and making development more flexible and targeted.

In addition, the channel has set up a developer communication community to facilitate everyone to share their experience and insights in using models and discuss technical problems together. This atmosphere of communication and cooperation can help developers quickly solve problems, stimulate innovative thinking, and promote common technological progress. Not only that, GitCode will continue to pay attention to the development of domestic large models, update the model version in the channel in a timely manner, provide the latest technical support, and the professional technical team is also on standby at any time to answer questions encountered by users during use, so that users have no worries.

GitCode and CodeArts provide a stable model download service, allowing developers to quickly obtain high-quality open source model resources. These models have undergone a strict open source governance process to ensure their security and compliance. At the same time, CodeArts' full-stack self-developed technology system and efficient code hosting capabilities provide solid technical support for model storage and distribution. GitCode has also integrated the world's core open source technology stack resources to further enrich the model library, helping developers to quickly develop and innovate in the intelligent era.

This marks the deep integration of domestic computing power and advanced AI models. Relying on the powerful computing power of Huawei Cloud Ascend, the DeepSeek Space service beta version can provide efficient reasoning performance, with performance comparable to high-end GPUs. Developers can use DeepSeek models for reasoning tasks more conveniently and enjoy the efficient computing power brought by Ascend GPUs.

We will make full use of the powerful computing power of Ascend GPU to provide users with efficient, stable and high-performance reasoning capabilities. The specific application process and usage guide will be announced after the service is launched, so stay tuned!

On the road to technological independence and AI power, GitCode, as a company rooted in China, will always stick to its mission and continue to promote the ecological construction of domestic large models. We believe that through continuous technological innovation and service optimization, GitCode can contribute more to the development of China's AI industry.