DeepSeek Launches V4 Pro Model, Raises API Pricing

Chinese artificial intelligence startup DeepSeek has officially released its V4 Pro model, moving the flagship system out of preview and into general availability while simultaneously announcing a significant restructuring of its API pricing. The new build, designated V4Pro0813, became available on August 13, 2026, across the company's app, web interface, and API, marking a major update as the firm seeks to maintain its competitive edge in a crowded market.

The release represents the general availability version of a model that had been in preview since April. DeepSeek has positioned V4 Pro as a powerful tool for complex, multi-step workflows, featuring a one-million-token context window that allows the system to process approximately 750,000 words in a single prompt. The company's published benchmark results show the new model significantly outperforming its April preview across a range of agent-focused tests, including terminal operation, code engineering, and tool use. According to evaluation data from Artificial Analysis, V4 Pro scored 53 on its Intelligence Index, placing it on par with Zhipu AI's GLM5.2 but behind models like Moonshot AI's Kimi K3 and OpenAI's GPT5.6 series.