Update README.md

README.md

@@ -138,6 +138,24 @@ vllm serve Qwen/Qwen3-Coder-Next --port 8000 --tensor-parallel-size 2 --enable-a
 > The default context length is 256K. Consider reducing the context length to a smaller value, e.g., `32768`, if the server fails to start.
 
 
+### KTransformers
+
+[KTransformers](https://github.com/kvcache-ai/ktransformers) is a CPU-GPU heterogeneous inference engine for large language models.
+KTransformers could be used to launch an OpenAI-compatible API service, with one single GPU.
+
+`KTransformers` can be installed and run followed by its [Qwen3-Coder-Next Tutorial](https://github.com/kvcache-ai/ktransformers/blob/main/doc/en/kt-kernel/Qwen3-Coder-Next-Tutorial.md).
+
+See [its documentation](https://github.com/kvcache-ai/ktransformers/blob/main/kt-kernel/README.md) for more details.
+
+The following command can be used to create an API endpoint at `http://localhost:30000/v1` with maximum context length 256K tokens on 1 GPU.
+```shell
+kt run Qwen3-Coder-Next
+```
+
+> [!Note]
+> The default context length is 256K. Consider reducing the context length to a smaller value, e.g., `32768`, if the server fails to start.
+
+
 ## Agentic Coding
 
 Qwen3-Coder-Next excels in tool calling capabilities.