DeepSeek Lifts AI Model Prices Fourfold
Even after the increase, it still charges far less than U.S. AI players and Chinese peers
- China’s DeepSeek is raising prices for its flagship V4 artificial intelligence models, with the increases taking effect on Aug. 16.
- New peak-hour pricing for the DeepSeek-v4-pro model will rise to $3.96 per 1 million output tokens, up from $0.87 previously.
- Even after the price hike, DeepSeek’s rates remain far below those of its U.S. and Chinese rivals.
China’s DeepSeek is hiking prices for its flagship V4 models amid fierce competition among artificial intelligence players, though its latest prices remain well below its rivals’.
New peak-hour pricing will be over four times higher than current levels, according to a post on its official website on Thursday. The increase will take effect on Aug. 16.
For DeepSeek-v4-pro model, users will pay $3.96 for 1 million output tokens during peak hours, and half that during off-peak hours. That’s up from $0.87 per million tokens previously. Tokens are the units of text that AI model process, and the basis on which they charge for their services.
The DeepSeek-v4-flash model will charge $1.32 per 1 million output tokens during peak hours, and half that during off-peak hours, the post said.
Even after the increase, DeepSeek still charges far less than U.S. AI players and Chinese peers. Moonshot AI’s Kimi K3, for example, charges $15.00 per 1 million output tokens.
DeepSeek is in talks with prospective investors for a new round of fundraising after raising more than $7.4 billion in June. The company is preparing to list shares in Shanghai as early as the second quarter of next year, the Wall Street Journal reported.