It is very common to use ChatGPT to search for different results on the web, although in many cases it is convenient to use a local assistant to access all the offline tools.
One of the best tools to achieve this is Ollama, an open source application that allows users to test different large natural language models (LLM) without having to access the internet.
For others, such as ChatGPT by OpenAI and Gemini by Google, you will have to access the service online, unless you configure them locally, something more complicated and not suitable for all computers.
Unlike those mentioned, Ollama allows you to directly choose the model you prefer from the same conversation with the assistant, with others that are not exclusively GPTbut also Gemma, Mistral or DeepSeek.
Best of all, it is not even necessary to register and offer your email, since the application can work correctly with access to all modelsas long as you don’t need to search the internet.
We tell you how you can use Ollama, an open source tool, to create your own ChatGPT assistant.
How to create an offline ChatGPT with Ollama
To start using Ollama, you simply have to access its official page from this link and choose the operating system of your computerwhich includes macOS, Windows and Linux.
The first thing you have to keep in mind is that at no time will you be asked to create an account and, although you can use it with a connection to a profile associated with your email to browse, the process to get the offline GPT model.
One of the most important issues that you have to take into account is the complete list of models available in Ollama, to know which one best suits to what you want; In case you need it to remember your conversations, you will have to change it in the settings.
This section will allow the LLM to acquire much more context, especially if you need to use it as a completely personal assistant.
Something to keep in mind is that the Ollama application itself weighs just over 1 gigabyte, to which you have to add considerably more storage available if you need to run an LLM locally.
For example, in the case that I have used, the gpt-oss:120b model, one of the most powerful you can try and that accumulate 3.3 million downloads, the total weight is 65 gigabytes. In this case, you always have to check the characteristics of the model you are going to test.
Ollama’s own website has a complete list of all modelswith the options available according to the number of parameters of each LLM, in addition to the last update date or the possibility of using it online.
In case you need the most powerful gpt model, like the one I mentioned, you simply have to access its page within the list and copy the first message that appears on the page, with the copy to clipboard symbol shortcutand paste it in Ollama.
From here, the tool will begin to download the complete language so that works locallyso you will need to have the space it takes up or, otherwise, you will not be able to use it.
This is the easiest way to use a model, although it is best to open the command from the terminal of your operating systemwith the same message that Ollama shares so you can copy and paste directly. With this, you will also be able to access cloud or local models.
As you can see, Ollama is a very powerful tool that brings together many LLMs, so can be a great companion If you usually try different ones and want to have them all in the same place, also available offline.