Local AI runtimes and tools

A source-labeled directory of local AI runtimes, browser UIs, and document-chat tools.

Local AI Guide is documentation-backed, not a benchmark lab. Records show source-reviewed facts, conservative estimates, and unsupported areas separately.

Understand the tool types

This directory explains the difference between local runtimes, desktop apps, browser interfaces, RAG/document apps, and lower-level libraries.

Do not treat all tools as equivalent. A runtime, GUI, RAG app, and self-hosted web UI solve different problems and carry different privacy risks.

Runtime, app, UI, or document workflow

Category	Meaning	Examples
Runtime	Runs models locally and exposes CLI/API behavior.	Ollama, llama.cpp
Desktop app	Gives users a GUI for downloading and chatting with models.	LM Studio, Jan, GPT4All
Browser UI	Web interface layered over local or cloud providers.	Open WebUI
RAG/document app	Focuses on document ingestion, retrieval, and workspace workflows.	AnythingLLM

Official documentation reviewed

Ollama

Ollama is a local model runner used to pull, run, and serve models locally. It often acts as the backend for other local AI apps.

Ollama is the best first record to enrich for users who want a lightweight local model runner, command-line workflow, local API behavior, or a backend for Open WebUI. It should be described as a runtime/backend first, not as a full document-chat app by itself.

Official documentation reviewed

LM Studio

LM Studio is a local AI desktop app and developer stack. It can be used as a chat app, model manager, document-chat app, and local server.

LM Studio is the best first local AI app for many beginners because it gives users a desktop interface for finding, downloading, and chatting with local models. It should be framed as GUI-first, with optional developer features.

Official documentation reviewed

Open WebUI

Open WebUI is a self-hosted-style web interface that can connect to local providers such as Ollama and to cloud-compatible providers.

Open WebUI is the right record for users who want a browser-based interface over local or cloud providers. It is powerful, but it should not be presented as a one-click beginner app.

Official documentation reviewed

AnythingLLM

AnythingLLM is an AI workspace/document-chat category tool. It can be useful for RAG-style workflows, but it should be evaluated by provider choice and storage behavior.

AnythingLLM belongs in Local AI Stack as a document/workspace-oriented local AI option. It should be covered carefully because provider selection, embeddings, telemetry, and storage choices determine how private the workflow actually is.

Official documentation reviewed

Jan

Jan is a desktop local AI app category entry. It should not be elevated above better-researched launch tools until its current features, provider behavior, and privacy posture are reviewed.

Jan should be treated as a desktop local AI app to investigate after the Ollama/LM Studio launch cluster is live. Keep this page conservative until current source review is complete.

Official documentation reviewed

llama.cpp

llama.cpp is a lower-level local inference project often used directly by advanced users and indirectly by local AI apps.

llama.cpp is a technical local inference backend/library, not the easiest beginner app. It is essential background for local model formats and quantization, but the main beginner path should usually be Ollama or LM Studio.

Official documentation reviewed

GPT4All

GPT4All is a local AI app/project category entry. It belongs in the ecosystem map but needs current source review before strong recommendations.

GPT4All should be included as a local AI desktop-app reference, but not prioritized ahead of Ollama, LM Studio, Open WebUI, and AnythingLLM for the launch cluster.