Three reasons why DeepSeek’s new model matters

Chinese AI firm DeepSeek has unveiled a preview of its new flagship model, V4, marking a significant advancement in open-source artificial intelligence technology. The model is designed to handle much longer prompts than its predecessor, thanks to a novel architecture that improves efficiency in processing large volumes of text. Like previous DeepSeek models, V4 is open source, allowing developers and companies unrestricted access to download, use, and modify the technology. The release follows DeepSeek’s breakthrough R1 model, which gained global attention for its strong performance and efficiency despite limited training resources. V4 is available in two versions: V4-Pro, tailored for coding and complex tasks, and V4-Flash, a smaller, faster, and more cost-effective variant. Both versions feature reasoning modes that enable the model to break down and explain its problem-solving process step-by-step. Pricing for V4-Pro is set at $1.74 per million input tokens and $3.48 per million output tokens, significantly lower than comparable offerings from major AI companies like OpenAI and Anthropic. V4-Flash is even more affordable, costing approximately $0.14 per million input tokens and $0.28 per million output tokens, positioning it as one of the most economical high-performance models available. DeepSeek’s V4 release comes amid heightened scrutiny of the company, including personnel changes, delays in previous launches, and increased oversight from both U.S. and Chinese authorities. Despite these challenges, the model’s open-source nature and competitive pricing could democratize access to advanced AI capabilities, enabling a broader range of developers and businesses to integrate cutting-edge technology without prohibitive costs. While V4 is unlikely to replicate the disruptive impact of R1, it represents a meaningful step forward in making powerful AI tools more accessible and affordable globally.