DeepSeek's V4 AI Model Challenges Closed-Source Rivals

Chinese AI company DeepSeek released a preview of its hotly anticipated next-generation AI model V4 on Friday, stating that the open-source model can compete with leading closed-source systems from US rivals including Anthropic, Google, and OpenAI.

DeepSeek says V4 marks a major improvement over prior models, especially in coding, a capability that has become central to AI agents and helped drive the success of tools like ChatGPT Codex and Claude Code. The release is also a milestone for China's chip industry, with DeepSeek explicitly highlighting compatibility with domestic Huawei technology.

The release comes a year after DeepSeek announced its previous V3 model, which showcased significant progress in large language models and was noted for its efficiency and coding prowess. The V4 model is expected to further push the boundaries of what open-source AI can achieve, potentially democratizing access to advanced AI capabilities.

This development is particularly significant given the ongoing global competition in AI development. While US tech giants have dominated the headlines with their proprietary models, DeepSeek's open-source approach with V4 signals a strong challenge. The emphasis on coding capabilities is a key indicator of the direction AI is heading, with a growing focus on practical applications and the development of more sophisticated AI agents that can assist with complex tasks.

Furthermore, DeepSeek's explicit mention of compatibility with Huawei technology underscores the growing strength and independence of China's domestic tech ecosystem. As the global AI landscape continues to evolve, the V4 model's performance and adoption will be closely watched as a measure of progress in both open-source AI and China's technological ambitions.