DeepSeek will permanently maintain the 75% discount on its flagship V4-Pro model that was set to expire on May 31, the company said on Saturday. V4-Pro API output tokens now cost $0.87 per million, down from $3.48 at launch one month ago.
DeepSeek announced on Saturday that the 75% discount on its flagship V4-Pro model is now permanent, bringing API costs down to $0.0035 per million tokens and intensifying the price war with OpenAI, Anthropic and Google.
The Chinese AI startup had originally set the promotion to end on May 31, according to its pricing page. Instead, the company said in a statement that V4-Pro API pricing will remain at a quarter of launch rates indefinitely, with costs ranging from 0.025 to 6 yuan ($0.003625 to $0.87) per million tokens depending on the type of usage.
Claude charges 29 times more for the same output code
With the new pricing structure, V4-Pro output tokens cost $0.87 per million tokens. Claude Opus 4.7 costs $25 per million tokens. GPT-5.5 costs $30.
According to CloudZero, Google’s Gemini 2.5 Pro costs $12. Other lower-cost Western LLM programs, such as Claude Sonnet ($15) and GPT-4.1 ($8), are 9 to 17 times more.
An application producing 100 million output codes per month would cost about $87 on V4-Pro at the new rate, versus $2,500 on Opus 4.7 or $3,000 on GPT-5.5, per MindStudio.
As Cryptopolitan reported earlier When V4 was launched last month, DeepSeek said the Pro version would initially cost up to 12 times more than the lighter Flash version due to “limitations in edge computing capacity.”
The company added that prices will drop once Huawei Ascend 950 supernodes are shipped in larger quantities in the second half of 2026. The permanent cut arrived months before that target.
DeepSeek did not say whether Huawei chips made the cut possible
DeepSeek did not reveal whether the permanent price reduction is due to the increased supply of Huawei Ascend 950 chipsets.
Huawei’s AI chip sales have benefited from US export controls that prevent Nvidia from selling its most advanced semiconductors in China. Separate restrictions on chipmaking equipment have limited Huawei’s ability to scale Ascend production.
Huawei is targeting 750,000 shipments of AI chips in 2026, according to industry estimates, although its total production amounts to roughly 3% to 5% of Nvidia’s combined computing power.
Four Chinese chip companies (Huawei Ascend, Cambricon, Hygon Information, and Moore Threads) have confirmed that their devices work with the V4-Pro since launch day.
The V4 launch has already sparked a sell-off in Chinese AI stocks
In early April, when V4 was introduced, the stock prices of some Chinese AI companies fell on the Hong Kong Stock Exchange. Zhipu AI performance dropped by about 8%-9% while MiniMax saw a 7%-8% drop.
According to CNBC, Manycore Tech stock fell 9%. Chip makers’ shares rose against the trend.
DeepSeek’s pricing strategy has also sparked friction with Western competitors. Anthropic has previously accused the company of “distillation attacks” that incorrectly learn from Claude models, according to Engadget.
Last month, the White House accused China of stealing the intellectual property of American artificial intelligence laboratories. The Chinese Embassy in Washington rejected these allegations.
DeepSeek, meanwhile, is in talks to raise external funding for the first time, with Tencent and Alibaba among potential investors at a valuation of more than $20 billion. Per Cryptopolitan.
“Crypto bots and DeFi agents finally have decent prices.”
Agents that independently execute multi-step tasks can burn millions of tokens per session.
The daily cost of running a proxy like Claude Opus at $25 per million throughput codes is hundreds of dollars. DeepSeek V4-Pro $0.87 Under $40.
Cryptocurrency trading bots, DeFi monitoring agents, code reviews, and market analysis applications rely on inference costs that remain below a minimum where the value of the output is greater than the cost of computation.
DeepSeek’s announcement of its perpetual pricing enables a category of proxy apps to be viable on its platform, which is not possible on Western platforms at the current rate.
OpenAI has been reducing its prices throughout the past year. Anthropic released Cloud Haiku 4.5 for $1/$5 per million tokens. Google introduced Gemini Flash with an entry price of $0.30. However, none of them have been able to compete with DeepSeek’s pricing model.
The V4-Pro is not a budget model. It has 1.6 trillion parameters, a context window of 1 million tokens, and benchmarks approaching GPT-5.4 in math and reasoning tasks, according to a MindStudio analysis.
