DeepSeek launches V4 Pro at prices up to 14 times higher than V4 Flash

Chinese AI startup DeepSeek has officially launched its V4 Pro model, charging significantly more than its V4 Flash counterpart as it aims to offer a premium flagship product. The V4-Pro-0813 model is priced at $1.32 per million input tokens and $3.96 per million output tokens, roughly nine times and 14 times the input and output prices of V4 Flash, priced at $0.14 and $0.28 per million tokens respectively, according to benchmarking firm Artificial Analysis.

DeepSeek stated the V4-Pro-0813 significantly improves its AI agent capabilities and is accessible via its API, app, and web products. The company also announced upcoming increases in API pricing for both V4 Pro and V4 Flash models, along with the introduction of peak and off-peak pricing. This pricing strategy places a higher premium on DeepSeek's top model following its V4 Flash model, which unexpectedly outperformed an earlier V4 Pro preview version in multiple independent tests.

This rapid improvement in technology suggests DeepSeek has made swift advancements between the preview release and the official launch. Notably, independent evaluations now indicate the official V4 Pro release represents a substantial step-up. AI analysis gave the V4 Pro's reasoning version an intelligence index score of 53, compared to 40 for V4 Flash, combining assessments of nine features such as agentic tasks, tool use, coding, scientific reasoning, and long-context performance.

DeepSeek gained prominence in China after its R1 model went viral in early 2025, sparking discussions about the potential for building powerful AI systems at lower costs than US rivals. However, its lead has been challenged by Chinese competitors such as Moonshot AI, Zhipu AI, MiniMax, Alibaba, and ByteDance. Additionally, DeepSeek faces the challenge of sustaining its prominence as a business.

Reuters reported in July that the company is planning a new fundraising round with a valuation of around $74 billion, following a $7.4 billion financing round in June. This fundraising marks a shift for a company that had previously avoided external capital, reflecting the increasing costs of competing in AI, which involve substantial investments in computer chips, data centers, and specialized staff.

DeepSeek has also announced plans to double its workforce across various departments, including data-center and AI-agent teams, and to increase private hiring of chip-design engineers to develop its own AI chips, potentially reducing reliance on suppliers like Nvidia and Huawei.