Concepts

Concepts

LM Nexus is a local-first AI control center: one place to run models and keep the tools, project context, and runtime controls around them.

Local-first, not local-only

Local-first means local control is the default, not a restriction.

LM Nexus is designed to work with:

• local models;
• self-hosted runtimes;
• custom OpenAI-compatible endpoints;
• connected API providers.

You can keep a workflow fully local when that makes sense, or connect to an external provider when it does not. The important part is that the choice stays visible and under your control.

Three ways to use it

Simple mode

Choose or download a model, click Run, and start working through the UI. You should not need a terminal just to get the first model running.

Expert mode

Open the runtime controls, change launch settings, inspect logs, copy endpoints, or connect directly to the server when you need lower-level control.

Workbench mode

Use the model as part of a larger workspace with agents, files, notes, knowledge, terminals, Git, logs, and other tools.

These are not separate products. They are different levels of control over the same environment.

What LM Nexus is not

LM Nexus is not just a chat UI. Chat is one way to work with a model, not the boundary of the product.

It is not just a model runner. Running a model matters, but so do the files, tools, providers, logs, and project state around it.

It is not just an agent framework. Agents are one part of the workspace and use the same models, tools, and project material as the rest of LM Nexus.

And it is not trying to present itself as an “AI OS.” The simpler description is the better one: an open-source, local-first AI control center and workbench.