en.Wedoany.com Reported - On July 31, DeepSeek officially announced the public beta of the DeepSeek-V4-Flash official API. The National Supercomputing Internet has officially launched the DeepSeek-V4-Flash official version (DeepSeek-V4-Flash-0731) model API calling service and model files.
DeepSeek-V4-Flash-0731 has undergone extensive post-training, significantly enhancing its agent capabilities and instruction-following abilities. In multiple benchmark tests, the DeepSeek-V4-Flash official version delivers performance comparable to the strongest current closed-source models.
The Supercomputing Internet now offers the DeepSeek-V4-Flash-0731 model API calling service to enterprises and developers. Users can access it with one click without cumbersome environment configuration, providing high-performance, secure, reliable, and affordable full-scenario AI development support for users across industries.
Users can log in to the National Supercomputing Internet official website (www.scnet.cn) on PC, navigate to the "Model Services" section via the "Services" dropdown menu on the homepage, and reach the model API calling page with one click.
The Supercomputing Internet platform provides an integrated "computing power + models" delivery service. Enterprises and developers can log in to the Supercomputing Internet AI community to download the DeepSeek-V4-Flash-0731 model immediately, and leverage the national-level computing infrastructure and secure, trusted Notebook development environment to complete the full development workflow—model deployment, fine-tuning, adaptation optimization, and more—in one stop.
With its core nodes officially operational, the National Supercomputing Internet has become the country's first 100,000-card-scale super-intelligent integrated computing resource pool. Leveraging the national integrated computing network, it provides massive computing resources for large model development, scientific agents, and other scenarios to supercomputing and intelligent computing users nationwide. In addition to abundant computing resources, the National Supercomputing Internet also offers MaaS services simultaneously. Whether developing AI applications, calling model interfaces, or building agent workflows, users can quickly access everything through a unified entry point, streamlining the complex processes of environment configuration, model adaptation, and computing resource scheduling.
To date, the Supercomputing Internet AI community has aggregated over 1,700 popular open-source large models from both domestic and international sources, covering mainstream series such as DeepSeek, GLM, Qwen, and Kimi. To further meet diverse calling needs, the platform has fully launched API interfaces for multiple domestic open-source large models, including DeepSeek-V4, MiniMax-M3, and Kimi-K3, deeply supporting core business scenarios such as code development, academic research, content creation, and intelligent Q&A.Whether you are an enterprise developer, researcher, or office user, you can enjoy a stable, efficient, and ready-to-use standardized calling experience, making AI application development easier.
