DeepSeek Unveils New AI Models Amid U.S. Concerns Over Intellectual Property Theft
Chinese AI lab DeepSeek has introduced two new versions of its large language model, DeepSeek V4, which are designed to improve upon its previous models. The V4 Flash and V4 Pro models utilize a mixture-of-experts approach, allowing for efficient processing with context windows of 1 million tokens. This enables the models to handle large codebases or documents effectively. The V4 Pro model boasts 1.6 trillion parameters, making it the largest open-weight model available, surpassing competitors like Moonshot AI's Kimi K 2.6. Despite these advancements, the models lag behind frontier models in knowledge tests, trailing OpenAI's GPT-5.4 and Google's Gemini 3.1 Pro by 3 to 6 months. The launch coincides with U.S. allegations against China for stealing intellectual property from American AI labs, with DeepSeek accused of 'distilling' AI models from companies like Anthropic and OpenAI. 
