DeepSeek Unveils Open-Source Tool to Help Cut AI Operating Costs

DeepSeek has released DeepInfer, an open-source inference framework designed to lower the compute costs and energy requirements of running large AI models. The Apache 2.0-licensed software, which currently powers the DeepSeek-V4.1-Flash model, is optimized for NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs. By providing businesses with a more efficient way to run models at scale, this move introduces further pricing pressure on Western AI providers that currently dominate the market.