LM Studio on Mac: install, load models and enable API calls

This tutorial explains how to install LM Studio on macOS and run models locally. It starts with the Mac download at lmstudio.ai, then covers installation, the first launch through Spotlight and the welcome screens. The speaker recommends around 16 GB of RAM and a current M-series chip; these are recommendations rather than a demonstrated minimum requirement.

The walkthrough downloads Google's Gemma, which appears as the suggested default model, and loads it into the chat interface. The speaker checks memory use in Activity Monitor on a Mac Studio M2 Max with 32 GB of RAM and reports stable usage with the selected Gemma model. The interface has chat sessions on the left and the conversation on the right.

Next, the guide uses model search to download a Qwen model described as having 9 billion parameters. To switch models, the speaker ejects Gemma to free RAM, adjusts the context-length slider and loads Qwen before starting a new chat. The tutorial explains that a larger token context consumes more memory.

The final section enables API access through the developer tab, selects a model and copies a curl example for a terminal test. OpenClaw appears as a possible integration, without a setup demonstration. The speaker also cautions that local models may be less capable than paid cloud models, but supplies no comparative tests.