DeepSeek V4's permanent price cut upends enterprise AI

DeepSeek permanently cut prices on its flagship V4 Pro and is promoting a cheaper V4 Flash model, both open-weight and MIT-licensed.

The company says hardware-software efficiency gains, especially cache optimizations, let it offer much lower input and output costs than Western rivals.

The move could commoditize high-volume enterprise AI tasks and intensify pressure on OpenAI and Anthropic’s infrastructure-heavy pricing models.

As companies look for cheaper alternatives, the market may split further between premium frontier models and low-cost commodity AI workloads.