DeepSeek cuts V4 Pro price 75%

DeepSeek said on May 24 it would keep a 75% discount on its V4 Pro model in place, extending a promotion that had been due to end on May 31. The move cuts V4 Pro output pricing to $0.87 per 1 million tokens, according to DeepSeek’s API pricing page, and puts the Chinese company far below published rates for leading U.S. models. The change lands as enterprise buyers weigh model capability, safety controls and operating cost more directly in procurement decisions, according to a May 24 analysis by Analytics Insight. It also adds a fresh pricing jolt to a market where OpenAI, Anthropic and Google have been competing on both performance and distribution. ### How big is the cut in dollar terms? DeepSeek’s API documentation lists V4 Pro at $0.435 per 1 million input tokens and $0.87 per 1 million output tokens after the discount becomes the official price. The same page says the 75% promotion had been set to expire on May 31 at 15:59 UTC before the company decided to keep the lower rate. (api-docs.deepseek.com) The Next Web reported on May 24 that the permanent discount takes V4 Pro pricing to a fraction of rivals such as GPT-5 and Gemini. MSN, citing the company’s announcement, reported that the output-token price fell from $3 to $0.87 per million. ### How does that compare with OpenAI’s published pricing? (api-docs.deepseek.com) OpenAI’s API pricing page lists GPT-5.5 at $30 per 1 million output tokens and $5 per 1 million input tokens. On that comparison, DeepSeek’s published V4 Pro output rate is about 97% lower than GPT-5.5’s listed output price. The gap matters most for workloads that generate long answers, code completions or agent traces, where output-token charges can dominate total cost. (thenextweb.com) DeepSeek’s own pricing page also lists a sharply reduced cached-input rate, which can further lower bills for repeated prompts and production traffic. ### Why does this matter for enterprise buyers now? (openai.com) Analytics Insight said on May 24 that enterprise adoption in 2026 is being shaped by pricing, safety posture, performance and control, rather than benchmark scores alone. The publication cited Ramp’s AI Index as showing Anthropic at 34.4% of paid business adoption in April 2026 versus OpenAI at 32.3%, while saying Gartner projects global AI model spending to approach $33 billion in 2026. (api-docs.deepseek.com) Those figures help explain why a model-price cut can affect more than developer experimentation. Procurement teams buying copilots, internal search, coding tools and customer-service agents often compare recurring token costs across vendors before expanding deployments, especially when multiple models are considered good enough for a given task. That is an inference from the pricing and enterprise-adoption data, not a direct statement from the companies. (analyticsinsight.net) ### Is this only about price, or also about competitive positioning? The Next Web described the move as an escalation in the AI price war. Its report said the lower V4 Pro rate undercuts flagship offerings from larger rivals at a time when model quality across vendors has narrowed enough for cost to become a stronger differentiator. (analyticsinsight.net) Analytics Insight framed 2026 as a live enterprise contest between Anthropic and OpenAI, with trust, governance and deployment fit shaping business adoption alongside raw capability. DeepSeek’s cut inserts a third pressure point into that contest: a much lower published cost base that buyers can use in negotiations even if they do not switch vendors outright. That last point is an inference based on the published price gap and enterprise comparison, not a quoted claim from Analytics Insight. (thenextweb.com) ### What should buyers watch next? May 31 at 15:59 UTC is the next dated milestone on DeepSeek’s pricing page, because that is when the promotional period had originally been set to end before the lower V4 Pro rate became the official price. OpenAI’s API pricing page and DeepSeek’s pricing documentation remain the clearest public references for direct cost comparisons, while enterprise-adoption trackers such as Ramp’s AI Index are likely to show whether lower pricing translates into broader paid usage. (analyticsinsight.net) (api-docs.deepseek.com)