In a low-key manner, DeepSeek launched the DeepSeek-V3-0324 model last night, branding it as a "minor iteration." However, its actual performance has been nothing short of impressive. The new version boasts significant enhancements in core functionalities, including code generation and front-end development, with certain capabilities now rivaling those of Claude 3.7 Sonnet. This update swiftly captured the attention of the global AI community, sparking lively discussions. The model's parameters have been expanded to 685 billion, and it now operates under the MIT license, facilitating commercial integration across a wide range of deployment channels. Improvements have been made in programming and technical skills, contextual comprehension, long-text processing, as well as technical infrastructure and performance, solidifying its strong competitive edge and industry influence.