China's National Supercomputing Network has launched the DeepSeek-V4-Flash official API, allowing one-click access for fast integration.

According to TrendForce Beating Monitor, the Chinese National Supercomputing Internet has officially launched the DeepSeek-V4-Flash final version (DeepSeek-V4-Flash-0731) model API call service and model files. DeepSeek-V4-Flash-0731, after extensive post-training work, has significantly enhanced the intelligence of the agent and the ability to follow instructions. In multiple benchmark tests, the performance of the DeepSeek-V4-Flash final version is comparable to the current strongest closed-source models.

The Supercomputing Internet is now offering the DeepSeek-V4-Flash-0731 model API call service to enterprises and developers. Users can quickly call the API without the need for cumbersome environment configuration through one-click access. Users can log in to the National Supercomputing Internet official website on the PC side, enter the "Model Service" section through the homepage's "Services" dropdown menu, and directly access the model API call page with one click.

Currently, the Chinese Supercomputing Internet AI community has gathered more than 1700 popular open-source large models, covering mainstream series such as DeepSeek, GLM, Qwen, Kimi, etc., and has fully launched multiple domestic open-source large model API interfaces such as DeepSeek-V4, MiniMax-M3, Kimi-K3, and more.