On October 20, 2026, DeepSeek announced the open-sourcing of its cloud service inference framework, DeepInfer, along with its production-grade tokenization component, deepseek-recipe, under the Apache 2.0 license. This framework supports high-throughput, low-latency large-scale inference services, capable of processing 155 trillion Tokens per day in a single cluster, generating approximately $38 in revenue per GPU card per hour. DeepInfer is compatible with both NVIDIA and Huawei Ascend platforms, supporting distributed inference and a variety of advanced features. The simultaneously open-sourced deepseek-recipe component unifies API protocols, optimizes long-text tokenization, and enhances system usability. DeepSeek's move aims to lower the barriers to AI adoption and promote the widespread use of AI technology. Its open-source initiative has been widely praised as a significant step toward democratizing AI.