Report
DeepSeek open-sources DeepInfer, but release lacks pieces of a complete inference service
DeepSeek says the framework powers DeepSeek-V4.1-Flash; its repository provides offline inference but omits components needed for a full service.
TLDR
DeepSeek released DeepInfer under the Apache License 2.0 license and says it powers DeepSeek-V4.1-Flash at scale. The repository lists NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs as supported hardware. Its release provides offline inference, but a KV cache system, disaggregation and load balancing are among the pieces it lists as necessary for a complete inference service. DeepSeek hopes sharing the framework will lower compute costs and energy use.
Combined views
33.2K
7 Sources, first seen ago
