DeepSeek slashes flagship AI model prices by 75 per cent in permanent shift
The new pricing structure drops costs to a quarter of the original rate, targeting enterprise users and intensifying competition with OpenAI and Google.
Chinese artificial intelligence startup DeepSeek has permanently reduced the price of its flagship DeepSeek V4 Pro model by 75 per cent. The move finalises a promotional discount that was previously scheduled to expire on 31 May 2026, effectively locking in lower costs for users indefinitely.
According to the company’s website, the new pricing structure ranges from $0.003625 to $0.87 per million tokens. This represents a significant drop from the previous rate of $0.0145 to $3.48 per million tokens, bringing the cost down to a quarter of the original level.
The decision comes one month after DeepSeek released its V4 models, including the Pro and Flash variants. The company marketed these releases as ushering in an "era of cost-effective 1M context length," aiming to establish the V4 Pro as a highly economical option for developers and businesses processing large data volumes.
By maintaining these reduced rates, DeepSeek intends to offer substantial savings for enterprise accounts and power users who process millions of tokens daily. The strategy appears designed to position the model as a direct, cost-effective alternative to major competitors, specifically citing OpenAI’s GPT-5 and Google’s Gemini 3.5 Flash.
The aggressive pricing strategy may provoke responses from established rivals. Anthropic, for instance, has previously accused DeepSeek of conducting "distillation attacks" to improperly learn from its Claude models. The permanent price cut underscores the intensifying competition in the generative AI market, where cost efficiency is becoming a critical differentiator for institutional adoption.
