DeepSeek releases DeepInfer inference engine under Apache 2.0 license
In brief
- DeepSeek releases DeepInfer, the inference engine for V4.1-Flash, under Apache 2.0 license.
- Limited compatibility with NVIDIA SM100 GPUs and Huawei Ascend 950 processors.
- Advanced features like KV caching remain internal.
DeepSeek publicly released its DeepInfer inference engine on October 20, 2026, under the Apache 2.0 license. According to the announcement, it supports NVIDIA SM100 GPUs and Huawei Ascend 950 processors but is limited to offline inference. DeepSeek states that features such as KV caching or load balancing between replicas remain internal and that full prefill/decode separation is not included.
This summary gives you the essentials. The full article, its figures and quotes are on ActuIA.
Read the original articleAI-generated summary of the source, checked by rules. How we write · Report an error