China's DeepSeek to raise AI model fees for developers

Chinese artificial intelligence company DeepSeek announced Thursday it plans to significantly raise its API access fees for developers and businesses, just days after launching V4-Flash, a new version geared towards programming tasks and autonomous tool use.

The company informed its users that it intends to "significantly" increase the prices for its API services, according to the Chinese media outlet National Business Daily. DeepSeek's V4-Flash is currently priced at $0.14 per million input tokens and $0.28 per million output tokens, the basic units of text processed by language models.

APIs, or application programming interfaces, serve as the channel through which developers and companies can connect their own applications, websites, or internal services to an AI model. This integration allows DeepSeek's capabilities to be embedded in commercial products, programming assistants, customer service systems, analysis tools, and services handling large volumes of queries.

The announcement comes after DeepSeek enabled API access to V4-Flash last week. This new version boasts enhanced agent capabilities, meaning systems that can perform tasks step-by-step, utilize tools, or advance programming work with reduced human intervention. Chinese media have highlighted V4-Flash as a faster and more cost-effective model to run compared to larger alternatives, featuring a structure designed to minimize resource consumption.

The planned price increase marks a new monetization phase for DeepSeek, which gained global recognition in early 2025 with its R1 model by offering high-performance AI at a low cost. The company's progress aligns with a trend among several Chinese tech firms to develop open-source models, allowing developers and businesses to download, adapt, and deploy AI systems within their own services. This approach contrasts with US firms like OpenAI and Anthropic, which have favored more restricted access models. The move occurs amid intense competition with other Chinese companies, including Alibaba, Tencent, Zhipu, and Moonshot, to develop more efficient models, in a sector heavily influenced by US restrictions on China's access to advanced AI chips.