DeepSeek unveiled its V4 Pro and Flash AI models, boasting 1 million token context length for seamless multi-turn conversations. The V4 Pro (1.6T total params, 49B active) claims parity with closed-source rivals, while V4 Flash (284B total, 13B active) prioritizes speed without sacrificing reasoning. Both models are fully open-source, enabling customization for researchers and developers.

The 1 million context window—double OpenAI's GPT-5.5's upper limit—allows the AI to maintain coherence in extended dialogues, a critical upgrade for complex tasks. DeepSeek asserts V4 Pro surpasses Gemini-3.1-Pro in "rich world knowledge" and matches closed-source models in reasoning benchmarks. V4 Flash, though less powerful, handles simple Agent tasks comparably to its Pro sibling.

Despite its technical strides, U.S. federal agencies and South Korean authorities have restricted DeepSeek due to national security and privacy concerns. The bans follow its meteoric rise as the top free iOS app in the U.S., underscoring geopolitical tensions around Chinese AI innovation.

By open-sourcing V4, DeepSeek aims to democratize access to high-performance AI, though debates persist about transparency and data ethics. The move intensifies competition with U.S. firms, potentially reshaping the AI landscape.