China’s DeepSeek prices new V4 AI model at 97% below OpenAI’s GPT-5.5
Chinese AI company DeepSeek has significantly reduced the pricing of its artificial intelligence models, including its new V4. Effective immediately and permanently, DeepSeek's API users will see input costs for previously processed context drop to one-tenth of the original price, with a minimum input cost of approximately $0.14 per million tokens.

Briefing Summary
AI-generatedChinese AI company DeepSeek has significantly reduced the pricing of its artificial intelligence models, including its new V4. Effective immediately and permanently, DeepSeek's API users will see input costs for previously processed context drop to one-tenth of the original price, with a minimum input cost of approximately $0.14 per million tokens. This pricing strategy makes DeepSeek's V4 model 97% cheaper than comparable OpenAI products. The move is expected to intensify competition within the AI market, potentially initiating a price war among major players.
Article analysis
Model · rule-basedKey claims
5 extractedThe price cuts for DeepSeek models are effective immediately and will be permanent.
The minimum input cost for DeepSeek API users is approximately US$0.14 per million tokens.
DeepSeek reduced prices for 'input cache hits' for API users to one-tenth of the original level.
DeepSeek priced its V4 AI model at 97 per cent less than OpenAI products.
DeepSeek's price cuts could potentially trigger a price war in the highly competitive AI market.