It’s a myth that you need super-powerful cloud computers to run AI services. In fact, your everyday PC or Mac can easily run apps that quickly spit out AI-generated text, potentially saving you from paying $20 a month for services such as ChatGPT Plus.
These apps are free and simple to use. They allow you to choose from a variety of large language models (LLMs) that have specific strengths, such as generating text responses or writing code.
They might not have the full capabilities of ChatGPT—they can’t generate images, for example, nor can they interrogate files or documents. But they can perform a huge number of AI tasks quickly, such as writing emails and text for a website, or explaining tricky concepts in plain English.
What’s more, they do all this without sending your personal data to the cloud, and without charging you a cent. You don’t even need to be connected to the internet to run them, either.
Here’s how to run two hugely powerful AI apps for free on your PC or Mac.
Jan
Jan is a super-friendly AI app that runs on Windows, Mac and Linux PCs.
Once you’ve downloaded it on to your computer, the first thing you must do is decide which AI model (or LLM) you want to use with the app. You can run multiple models and they’re mostly free to download and use. (You’ll also see ChatGPT on the list of available models, but bear in mind you’ll need paid-for access to the OpenAI API to make ChatGPT work in Jan.)
Which model should you pick? The Trinity model is a good choice for general purpose tasks such as writing copy or explaining concepts. The Gemma model is also built from the same AI technology that powers Google’s Gemini AI and is worth experimenting with.
The list of AI models in Jan reveals how much storage space they will take on your computer. The smaller models might take only a gigabyte or two; the more complex models can take 30 or 40GB. Jan will also warn you if you’ve not got enough computer memory (RAM) to run certain models, but most PCs and laptops should be able to run the more basic models without any trouble.
Once you’ve picked and downloaded your model(s), you can click the chat bubble in the left-hand menu to start talking to your AI. In the right-hand pane there are a few settings you can tweak before beginning your chat.
The Model setting lets you choose which of the downloaded AI models to chat with. The Instructions field lets you tailor the responses you’ll get back from the AI assistant. For example, if you’re using the AI to help you write guide for children, you might add instructions such as “explain everything as if I’m an eight-year-old.”
Now you can just start chatting. One of the tasks that apps such as Jan are great at is writing business emails from bullet points, which is particularly useful if your language skills aren’t great or you have a condition such as dyslexia.
For example, I gave the Trinity model the following instruction: “You are helping me write business emails for a car repair business.”
I then told it to: “Write email to Mrs Taylor. Thank her for email. Say we can look at her car at 9am on Weds 27 Mar. Park in bay C and drop keys in box at reception. Will leave message on her cell when car ready to collect.”
Within a few seconds it had written a perfectly professional, five-paragraph email with all the key details, which can easily be copied and pasted into an email app using the icons that appear at the bottom of each response. It’s a great example of how a free AI tool can add real business value.
LM Studio
LM Studio is very similar to Jan. It lets you pick from a menu of downloadable AI models, chat with the AI, or even run your own ChatGPT-like AI server if you’ve got the technical chops to set one up.
The interface is more geeky than Jan’s, but that may well suit more advanced users, especially coders.
LM Studio has also been well established for a while now and has a very active community of users on Discord, although again the discussions tend to be at the more techy end of the scale.
Once you start chatting away in LM Studio (click the chat bubble in the left-hand menu to begin a new chat), you’ll find a range of settings in the right-hand pane, including the very useful hardware settings. These allow you to set how much of your computer’s resources are used to power the AI. If you want faster responses, move the slider to the right. If you find LM Studio is putting too much strain on your computer, move it to the left.
Again, LM Studio offers access to models that are suited to general chat/copywriting tasks or those devoted to coding.
In the Playground section of LM Studio (click on the joystick icon in the left-hand menu), you can even run multiple AI models simultaneously to see which gives the best results. I asked the Gemini and Llama models to write bullet-point instructions on how to remove an app in Windows 11.
Gemma gave me two different methods for doing it, although the single set of instructions provided by Llama were easier to follow. But with multiple AI models working for you at the same time, you can pick and choose which works best for you!