Run Google Gemma 4 Offline on Your PC Using LM Studio (Full Transcript)

[00:00:00] Speaker 1: This is Google's Jemma 4 running directly on my PC for free. In this video, I'll show you exactly how to set it up step by step. Here's what it looks like. I'll paste in some meeting notes and ask it to pull out the action items. And there they are, fast and clean. Jemma 4 is Google's open AI model, meaning that there are no subscriptions, your data stays on your machine, and it even works offline. I'm Kevin, and let's set it up. To get started, download and install LM Studio. You'll find a link right here at the bottom of the screen. Now, I'm using LM Studio because it's the easiest way to run these models on your PC, and you don't need to use the command line or any complicated setup. Once you finish installing LM Studio, over on the left-hand side, let's click on the search icon. And right up here at the very top, let's type in Jemma 4, and you'll see a few different model sizes. The smaller models are faster, and the larger models are more powerful. I'll use the 4B model. I find that it's a great balance and works well on most PCs. I'll click on this, and then over on the right-hand side, let's click on download. Of course, you can always try the other models later to compare the results. Once the download is all done, down below you'll see a notification, and we can now use it in a new chat. Let's do that. Let's start with something practical. I have some meeting notes, and I'd like it to pull out the different action items. So here I have my meeting notes, and right down below, here I'll ask it to extract the action items and assign an owner if possible. Now right down below, you can choose the model that you would like to run. Now currently, I only have one installed, but if you install others, this is where you could choose them. If you want more detailed reasoning, you can turn on what's called think mode. It's a little bit slower, but I find that it's more thorough. I'm going to leave it on. Then over on the right-hand side, let's run it. And there we go. It pulled out all the key tasks, it cleaned them up, and it even suggested owners based on the context. Once you get your result, you have a few different options right here. Here, if I scroll down to the bottom, you'll see all these different actions. You can regenerate to get a different answer. Over here, you can branch to explore a new direction without losing any of this content that you see. You could also copy the output and use it somewhere else. Right over here, you can edit the result directly on the screen, or if you just want to start over, you can also just delete the message. So far, we've just been looking at how this works with text, but you can also use it to understand images. Right over here, I have a photo of a whiteboard after a meeting, and I'd like to understand this better. So down below, I could click on this plus icon, and I could attach a file or attach an image, but my preferred approach is simply to drag and drop the image into the chat field. Now, I could type in my prompt, summarize this whiteboard, and list the key takeaways and action items. Let's run that. And just like that, it turns all of this messy information into something clear and structured. LM Studio also has a few other features that make it easier to stay organized. In the top left-hand corner, let's click on this icon, and here you can see all of your different chats. And one neat thing is you could even group them into folders. Over on the right-hand side, if you click on this icon, you can open up a split view. That way, you could see multiple chats side by side. Now, this is great if you want to compare results. Maybe you have multiple models installed and you want to compare them. Also, within a chat, you could click on this three-dot icon, and right down here, you have the option to add instructions. That way, it'll respond the way you want it to every time. As an example, you could tell it to, let's say, respond in a specific format. Like, maybe you only want bullet points, or maybe you want a table in the response. You have that power and control. Of course, this is what most people expect from AI today. The difference, though, is it all runs directly on your PC. No subscriptions, no limits, and best of all, your data stays private. Give it a try and let me know what you think. Thanks for watching.