DeepSeek makes AI price cut permanent
DeepSeek has made the temporary 75 percent discount on its V4 Pro model permanent. According to The Next Web, the price is dropping to a maximum of $0.87 per million output tokens. The promotion was originally scheduled to end on May 31, 2026.
With this move, DeepSeek is further intensifying price pressure in the AI market. TNW compares the new price to more expensive models from OpenAI, Anthropic, and Google. Such a price difference can have a significant impact, especially for companies that process millions of tokens daily.
DeepSeek positions V4 Pro as a model for long context and cost-sensitive applications. According to reports, the model supports a context window of 1 million tokens. This makes it attractive for document analysis, codebase understanding, legal review, and long conversations, where token costs quickly escalate.
The strategic objective seems clear: DeepSeek is prioritizing market share over maximizing revenue per token. If model quality is good enough, developers and companies can deploy cheaper models for many routine tasks and reserve more expensive frontier models for more complex or sensitive tasks.
At the same time, uncertainty remains. For European and American companies, compliance, data security, geopolitical risks, and transparency regarding training data play a role. TNW also mentions the unresolved accusation by Anthropic that DeepSeek trained on Claude output via distillation. DeepSeek has not publicly refuted that claim in detail.
This is important for the AI market because token prices are increasingly moving towards a commodity economy. If DeepSeek continues to lower the bottom of the price curve, Western providers must choose: lower prices further, add more value through platform features, or focus more strongly on enterprise trust and regulation.