A 75% permanent price cut now applies to DeepSeek's flagship V4-Pro model. The company did not specify if this move stems from increased supply of Huawei Ascend 950 chips used for performance optimization. This aggressive pricing pressures competitors on cost-per-token. Developers can now scale high-performance inference with significantly lower overhead.