DeepSeek于8月6日发布公告，宣布计划近期整体上调API服务定价，预计涨幅较大[1][2]。目前该公司按输入Token和输出Token分别计费，根据缓存命中情况价格存在差异[1]。
DeepSeek的主要模型包括DeepSeek-V4-Flash和DeepSeek-V4-Pro[1]。其中，DeepSeek-V4-Flash的当前定价为：缓存命中情况下输入0.02元/百万Token，缓存未命中输入1元/百万Token，输出2元/百万Token[1]；DeepSeek-V4-Pro的当前定价为：缓存命中输入0.025元/百万Token，缓存未命中输入3元/百万Token，输出6元/百万Token[1]。业界此前普遍认为DeepSeek的API定价偏低[2]。
DeepSeek has announced plans to substantially raise the pricing for its API services in the near term [1][2]. The announcement was made on August 6, 2024 [1]. The company has indicated the adjustment will be considerable, with specific details to follow in an official notification [2].
Currently, DeepSeek charges separately for input and output tokens, with rates varying based on cache hit status across its main models [1]. For DeepSeek-V4-Flash, the existing rates are 0.02 yuan per million tokens for cached input, 1 yuan per million tokens for uncached input, and 2 yuan per million tokens for output [1]. DeepSeek-V4-Pro is priced at 0.025 yuan per million tokens for cached input, 3 yuan per million tokens for uncached input, and 6 yuan per million tokens for output [1]. Industry observers have previously noted that DeepSeek's API pricing was relatively low [2].
