LocaLM
Offline CLI tool to run local LLMs for chat, coding, image/music/video, RAG, memory and voice.
Tested on Linux
Use it
Linux x86_64
You need first
- Runtimes:
python>=3.12,<3.13 - A GPU is optional
Install localm via uv tool
uv tool install localmProvision CPU llama backend
export LOCALM_HOME="$HOME"/localm-data && localm setup-llama --backend cpu
Where the README and reality differ
- README: docs say API served over HTTPS with built-in certificate
What worked: plain `http://127.0.0.1:8642/health` and `/v1/models` work; `https://` fails with curl exit 35 SSL connect error
Verified in pickfit: the commands exactly as its sandbox ran them
For the record only: they name the sandbox's home, /home/sbx.
uv tool install localm 2>&1export LOCALM_HOME=/home/sbx/localm-data && localm setup-llama --backend cpu 2>&1localm --version 2>&1export LOCALM_HOME=/home/sbx/localm-data && mkdir -p "$LOCALM_HOME" && localm doctor 2>&1Check that it works
localm --version 2>&1The output should contain “localm, version 0.3.2”.
First thing to try
export LOCALM_HOME="$HOME"/localm-data && mkdir -p "$LOCALM_HOME" && localm doctor 2>&1Run localm doctor first to check your setup.
What it does
- Chat with local models and see replies as they streamTested
How to use it
- Add a downloaded GGUF model with
localm add <model.gguf>. - Chat with it using
localm run <model> --prompt "<your text>". - To serve it over an OpenAI-style API, run
localm serve <model> --port <port>and send requests tohttp://127.0.0.1:<port>/v1/chat/completions.
- Add a downloaded GGUF model with
- Run a coding agent that edits filesNot tested
- Generate images, video and music locallyNot tested
- Chat with your own files and get citationsNot tested
How far pickfit got
Linux x86_64
- Installs
- Starts
- 1 of 4 features work
- Works offline
Windows: the project says it runs there; pickfit hasn’t tried it.
Good for
- chat with a local LLM offline in terminal
- run autonomous coding agent to edit files and run tests
- generate images, music and video locally
- chat against local files and folders with citations
- dictate with local speech-to-text and listen to replies
An alternative to LM Studio and Ollama.
How pickfit tested it
Each run starts from a clean machine and follows the steps above. How pickfit verifies
Linux x86_64
It produces model output
- Replayed from scratch
- Works offline
- Checked on 30 Oct 2026
- Version 0.3.2
Differences the test machine hit
- README: `pip install localm` then `localm setup-llama`
In the test: use isolated `uv tool install localm`; set LOCALM_HOME=/home/sbx/localm-data and run `localm setup-llama --backend cpu` before any run/serve - README: SET LOCALM_HOME once in shell profile
In the test: `export LOCALM_HOME=...` on every command or version/doctor warn about contained default inside the uv tool venv