Player not loading? Watch on YouTube
Kevin Stratvert walks through setting up Google's Gemma 4 in LM Studio to run models locally on a PC without using the command line. He searches the model catalog, downloads the 4B model, and opens a new chat. He describes that size as a practical balance for most PCs, but gives no specific RAM or GPU requirements.
The first example asks Gemma 4 to extract action items from meeting notes and assign owners where the context allows. Stratvert enables think mode, which he says produces more thorough reasoning at the cost of speed. He then explains how to regenerate an answer, branch a conversation, copy or edit the output, and delete a message.
For image analysis, he drags a whiteboard photo into the chat and asks for a summary with takeaways and action items. The tutorial also covers chat folders, a split view for comparing conversations or models, and chat instructions for requesting formats such as bullet points or tables.
Stratvert says the setup works offline and keeps data on the user's machine. These are claims made in the tutorial; it does not include a privacy audit or hardware benchmarks.
