What is LM Nexus

LM Nexus is an open-source, local-first AI control center. You can run a model, talk to it, give it tools, keep project files and notes nearby, and still see the runtime underneath.

The point is not to replace every AI app. It is to stop a local AI workflow from being split across half a dozen of them.

The problem

Local AI gets fragmented quickly. One app runs the model, another handles chat, a terminal keeps the server alive, notes live somewhere else, and logs or project files end up in yet another window.

That is manageable for a quick conversation. It gets painful once you add agents, longer-running work, approvals, devices, or project state that needs to survive beyond one chat.

LM Nexus brings those pieces into one workspace without pretending they are all the same thing.

How LM Nexus fits together

Think of LM Nexus as one workspace with several connected areas:

Models & Runtime — download models, connect providers, start runtimes, inspect logs and metrics, or use a raw endpoint directly.
Agents & Tools — give models tools and skills, with visible approvals and limits on what they can do.
Workspace — keep Files, Git, Terminal, Notes, Knowledge, logs, and project material beside the model.
Physical Systems — connect local devices and choose which actions LM Nexus or an agent can use.
Living Worlds — keep persistent simulated worlds with their own characters, relationships, timelines, lore, and state.

These areas can work together, but LM Nexus does not pour all of their data into one giant AI memory. Files stay files. Device state stays with the device layer. A world keeps its own state. The model gets access to what it needs through tools and context.

Local-first, not local-only

Local-first means local control is the default. You can run models on your own machine, keep project data local, and inspect the processes that are doing the work.

It does not mean LM Nexus only works with local models. You can also use self-hosted runtimes, custom OpenAI-compatible endpoints, and connected API providers.

The model can change without forcing the rest of your workspace to change with it.

Beyond chat

Chat is useful, but it should not become the database for your whole project.

LM Nexus lets a model work with files, tools, notes, runtime state, local services, and devices without stuffing all of that state into the conversation transcript. That is the idea behind Language models, beyond chat.

Modular on purpose

LM Nexus is modular so new features can add pages, tools, settings, and integrations without giving every part of the app unrestricted access to everything else.

That makes the system easier to understand and debug, and keeps it clearer which part of LM Nexus owns which data or action.

For a deeper technical view, continue with Core concepts or explore the Product tour.