Chinese AI startup DeepSeek has announced steep price cuts for its newly launched V4 model, offering developers discounts of up to 75 percent as it intensifies competition in the global artificial intelligence market.
The company said the promotional pricing for its flagship DeepSeek-V4-Pro model will remain in place until May 5, alongside broader cost reductions across its API ecosystem. Prices for input cache hits have been cut to just one-tenth of their previous levels, significantly lowering the cost of deploying AI applications.
The move comes shortly after DeepSeek previewed its V4 model, which is available in two variants: a high-performance Pro version and a lighter, more cost-efficient Flash version. The models have also been adapted to run on Huawei’s chip architecture, reflecting China’s push towards greater technological self-reliance.
According to the company, the Pro version delivers strong performance in world-knowledge benchmarks, trailing only Gemini-Pro-3.1, developed by Google, despite being positioned as an open-source alternative.
DeepSeek said the V4 models are particularly suited for AI agent-based applications, which are designed to perform complex, multi-step tasks beyond traditional chatbot capabilities, but typically require higher computing power.
The aggressive pricing strategy signals DeepSeek’s intent to rapidly expand its developer base and challenge established players by making advanced AI models more accessible and cost-effective.
The development also highlights a broader trend in the AI industry, where pricing, performance, and infrastructure compatibility are becoming key battlegrounds as companies race to scale adoption globally.
