LM Studio for Developers: API, MCP, Auth, Headless Mode & More

Published 16/06/2026

Description

In this video, I explore the developer side of LM Studio and show how to use it as more than just a chat app. If you're a developer working with local LLMs, Windows, Ubuntu, or a DGX Spark, this walkthrough covers the features that turn LM Studio into a real local AI backend. Contents: 0:00 - Introduction 0:12 - Description 1:09 - Developer Settings 4:09 - Developer Tab 12:21 - llmster 14:44 - LM Studio Api - List Models 15:54 - LM Studio Api - LOad Model 17:50 - LM Studio Api - Unload Model 18:48 - LM Studio Api - Download Model 22:17 - LM Studio Api - Chat 29:53 - Authentication 31:54 - OpenAI Compatible Api 35:59 - Antropic Compatible Api 37:07 - Conclusion You’ll see how to use the LM Studio REST API, the OpenAI-compatible API, and the Anthropic-compatible messages endpoint. I also walk through developer settings, server configuration, authentication, CORS, MCP integration, just-in-time model loading, model unload/download APIs, streaming responses, and how to manage models locally or on remote devices. We also compare how LM Studio behaves on Windows versus Ubuntu headless mode, including what “headless” really means, how the local server works, and why you may need a reverse proxy if you want HTTPS in front of LM Studio. Topics covered in this video: LM Studio developer tab and local server settings OpenAI-compatible HTTP API LM Studio native REST API Anthropic-compatible API support Model loading, unloading, and downloading through API calls Authentication with bearer tokens MCP server usage and per-request MCP configuration Streaming responses and response IDs Windows vs Ubuntu setup differences DGX Spark and remote model usage Why LM Studio can become the backbone of your local AI stack If you're building tools, automations, or internal AI apps with LM Studio, this video will help you understand what’s available under the hood and how to use it in a more infrastructure-friendly way. #LMStudio #LocalAI #AI