DeepSeek Cuts V4‑Pro Model Prices by 75%, Pressuring Western AI Vendors

Context Chinese startup DeepSeek announced a permanent price cut for its flagship V4‑Pro model just weeks after the V4 generation launch. The adjustment follows a broader trend of falling inference costs, a key barrier for enterprises scaling AI projects.

Key Facts - New pricing drops from $0.0145–$3.48 to $0.003625–$0.87 per million tokens, a 75% reduction. - Greyhound Research analyst Sanchit Vir Gogia says the cut reflects an efficiency gain: V4‑Pro uses roughly a quarter of the compute and a tenth of the memory of its predecessor for long‑context tasks. - Counterpoint Research’s Neil Shah notes V4‑Pro now matches Western models on complex math and reasoning while leading on openness and cost, though it lags in ecosystem support and global services. - Enterprises can achieve further savings by deploying the model on‑premises, turning previously uneconomical workloads—such as always‑on copilots or bulk document review—into viable options. - The price drop puts pressure on OpenAI, Anthropic and Google, whose token rates remain higher and may force a shift toward outcome‑based pricing or multi‑model strategies.

What It Means The aggressive pricing makes a high‑performing, open‑weight model financially attractive for businesses that have struggled with the high cost of AI inference. CIOs may now consider mixing models—using premium Western services for critical tasks and DeepSeek for volume work—to optimize spend. However, adopting a Chinese‑origin model raises concerns about data sovereignty, regulatory compliance and geopolitical risk, prompting careful risk assessments.

Watch for how major AI vendors respond—whether through price adjustments, bundled services or new value‑based pricing structures—as enterprises begin to test DeepSeek’s cost advantage at scale.