DeepSeek Launches New AI Model, Challenges U.S. Rivals

On Friday, Chinese startup DeepSeek released its highly anticipated AI model, DeepSeek V4, which boasts ultra-long context capabilities of up to one million words. The company announced two versions: DeepSeek V4 Pro, with 1.6 trillion parameters, and DeepSeek V4 Flash, a more efficient model with 284 billion parameters.

This release positions DeepSeek as a significant player in the open-source AI field, outperforming both top-tier closed-source models like Google's Gemini Pro 3. The launch has raised questions about data privacy and censorship, particularly regarding the model's refusal to answer sensitive questions, such as those related to the 1989 Tiananmen Square crackdown.

Chinese Premier Li Qiang emphasized the intensifying AI race between China and the United States, with U.S. officials accusing Chinese entities of efforts to steal American AI technology. The release follows DeepSeek's emergence in January last year and reflects a broader trend towards open-source AI development in China.