DeepSeek Slashes AI Model Fees Again, Escalating Competitive Pressure on Western Rivals

Chinese AI laboratory DeepSeek announced another significant price reduction for its DeepSeek-V4-Pro model just days after its initial release, intensifying competitive pressure on Western AI companies that are simultaneously launching more expensive offerings. The latest cut follows a 75% base discount with an additional 90% reduction for cache hits, bringing the effective cost to approximately $0.003625 per million input tokens for cached queries. The promotional pricing runs through May 5, though the cache discount has been made permanent.

The pricing structure distinguishes between cached and uncached queries, with uncached input tokens priced at $0.0145 per million tokens. DeepSeek senior researcher Victor Chen announced the promotional structure on social media platform X, framing the move as an opportunity for users to "go wild." The company released the DeepSeek-V4-Pro model on Saturday, immediately implementing the first price reduction before announcing the second cut within 48 hours. This marks a notably aggressive pricing strategy compared to competitors Anthropic, OpenAI, and Google, which are rolling out newer, costlier models.

The timing signals a strategic divergence between DeepSeek and Western AI firms that are raising prices on advanced capabilities while DeepSeek drives costs downward. Industry observers have noted the move could accelerate a broader realignment in AI pricing structures, potentially reshaping how companies monetize artificial intelligence development. DeepSeek's continued fee reductions place pressure on Western competitors to justify premium pricing as the Chinese laboratory expands its market positioning through aggressive cost advantages.