diff --git a/scicomp/local_llm b/scicomp/local_llm
new file mode 100644
index 000000000..7ba153b65
--- /dev/null
+++ b/scicomp/local_llm
@@ -0,0 +1,27 @@
+==================
+Running Local LLMs
+==================
+
+This page describes options for running large language models (LLMs) locally.
+Most users should use already available services :doc:`local-llm-web-apis`
+or run on Triton :doc:`python-environment-for-machine-learning`.
+
+Llama.cpp
+---------
+
+Llama.cpp is a simple and secure way of running LLMs locally. To install it,
+ - find the latest release from ``GitHub <https://github.com/ggml-org/llama.cpp/releases>`__`
+ - download the appropriate binary version for your operating system and architecture
+ - extract the archive
+
+You should be able to run the server directly from the extracted folder. Locate the executable
+called `llama-server`. You will need a model file, for example
+``https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct-GGUF/resolve/main/qwen2.5-1.5b-instruct-q4_k_m.gguf <https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct-GGUF/resolve/main/qwen2.5-1.5b-instruct-q4_k_m.gguf>`__`.
+
+``Huggingface <https://huggingface.co>`__` is a cood place to look for model files. Look for
+a gguf file, that is the binary version of the model.
+
+To run llama.cpp, run the command `llama-server -m qwen2.5-1.5b-instruct-q4_k_m.gguf --jinja`.
+See ``llama.cpp documentation <https://github.com/ggml-org/llama.cpp#llama-server>`__` for details.
+
+Llama.cpp has a default browser interface, which will open at 127.0.0.1:8080 by default.