DeepSeek reopens access to its API after three-week pause

After a three-week hiatus due to capacity constraints, Chinese AI startup DeepSeek has reopened its API for access. Customers can now top up credits to use the API, which allows developers to create apps and services using DeepSeek’s cloud-hosted AI models. However, server resources are still limited during peak hours, according to a company representative’s message on WeChat as reported by Bloomberg.

DeepSeek gained recognition earlier this year with the release of its R1 “reasoning” model, which rivals the performance of top models from OpenAI. This competition has led OpenAI to consider open-sourcing more of its technology and adjusting product releases to stay competitive.

In response to DeepSeek’s resurgence, domestic competitors in China are also stepping up their game. On the same day that DeepSeek resumed API top-ups, Alibaba, a major tech company, unveiled its new reasoning AI model, QwQ-Max, with plans to make it open-source.