DeepSeek open sources DeepInfer, the engine behind its V4.1 Flash service
DeepSeek released DeepInfer, the in-house inference framework it says powers DeepSeek-V4.1-Flash at scale, as open source. The company says sharing its production stack should help the community lower compute costs, reduce energy use and make AI more affordable. The GitHub release supports NVIDIA and Huawei Ascend hardware and notes that a complete serving setup still needs parts not included, such as a KV cache system. These details come from the post by @deepseek_ai and the project page.