Competition in the artificial intelligence (AI) market has escalated with DeepSeek announcing it will permanently slash prices for its flagship artificial intelligence tool by 75%.
The Chinese artificial intelligence startup said it would lower the price of its V4 Pro model to between 0.025 and 6 yuan per million tokens ($0.0035 to $0.83), depending on usage type, from 0.1 to 24 yuan per million tokens, which are units of text.
“After the extended offer ends the DeepSeek-V4-Pro API price will remain unchanged. It is official set to ¼ of the original price?” DeepSeek said in a post on X.
The announcement followed months of speculation about whether DeepSeek could sustain its aggressive pricing strategy.
The company is best known for building large language models and AI systems similar to Chat GPT from OpenAI, Claude from Anthropic, Gemini from Alphabet unit Google and others, focussing on high-performance, low-cost models.
It shocked the AI world last year when it released its V-series models like V3 and V4 Pro which offered frontier-level AI performance at a fraction of the normal price.
DeepSeek did not confirm whether the permanent price reduction was related to improved access to computing resources, including increased supply of Huawei’s Ascend 950 chips.
When the V4 was introduced, the Pro version was priced significantly higher than its lighter ‘Flash’ counterpart due to constraints in high-end compute capacity, but DeepSeek said pricing would fall once large-scale deployments of Huawei’s Ascend 950 supernodes became available in the second half of the year.



