Originally published at: How to connect Ollama to ONLYOFFICE and use its AI models when working on documents | ONLYOFFICE Blog
We continue our series of articles explaining how to connect and effectively use various AI models in ONLYOFFICE. Today, we will focus on Ollama, a platform designed to set up and run AI models locally, and bring to light how to integrate it with the ONLYOFFICE editors and use one of its models for text-related tasks. Keep reading for more information.
About Ollama
Ollama is an open-source project whose user-friendly platform allows you to install, set up and run a huge number of large language models (LLM) locally on your machine, whether it be Linux, Windows or macOS. The platform provides easy access to its library where you can choose one of the available models. Each model has a description of its capabilities and required parameters as well as instructions on how to run it within Ollama so you can easily find what you need depending on your goals and hardware configuration.
Currently, Ollama’s library includes a variety of popular AI models, such as Llama 3.3, DeepSeek-R1, Phi-4, Mistral, Gemma 2 and more, that can be used for different purposes. Programming, text and image generation, information processing, question answering, etc. — Ollama has everything to get you covered.
How to integrate Ollama and ONLYOFFICE
Let’s find out what you need to make the ONLYOFFICE and Ollama integration work. The process will be divided into several steps to make it easier for you.
Step 1. Install Ollama on your local machine
If you are eager to try one of the AI models offered by Ollama in ONLYOFFICE, you need to have Ollama installed on your local machine. On the official website, you can find and download the corresponding client for your OS. In our case, it’s Windows 10.
The installation process is intuitive, so the platform easily gets installed with a few clicks.
Step 2. Download and run the required AI model
After successful installation on Windows, Ollama will run in the background, and you can access it via Cmd.exe, PowerShell or your favourite terminal application.
If Ollama is running well, it’s time to install the model you need. Please note some powerful models need a lot of RAM and require a lot of free space on your HDD. Choose wisely taking into consideration the configuration of your hardware.
Let’s try something lightweight but quite powerful for modern-day tasks. Llama 3.2 is a good choice since it needs only 2 GB of free space. Run this model in the Ollama native application with the following command:
ollama run llama3.2
For other models, you can find the corresponding commands on the Ollama website.
Step 3. Install ONLYOFFICE Desktop Editors
Since Ollama is designed for running AI models locally, its integration is possible only with ONLYOFFICE Desktop Editors, a free desktop app for Windows, Linux and macOS. You can download it from our website here.
Step 4. Install the AI plugin
AI integration in ONLYOFFICE works via a special plugin that you need to install. No AI model is configured beforehand so you also need to do some configuration work.
Launch ONLYOFFICE Desktop Editors and open one of the editors by creating a new office file or uploading one from your device. After that, go to the Plugins tab, open the Plugin Manager and install the AI plugin by clicking the Install button.
After installation, click the Background plugins button on the Plugins tab and activate the AI plugin.
Step 5. Add Ollama and configure the plugin
Now everything is ready to connect Ollama to the ONLYOFFICE desktop app. Go to the AI tab and click the Settings icon. In the AI configuration window, click Edit AI models and then hit the Add button.
In the Provider section, choose Ollama, and you will see http://localhost:11434 as the URL parameter. This is the standard localhost on which the Ollama API will be served.
Now just select the model you have installed on your machine. In our case, it’s Ollama [llama3.2:latest]. To finish, click OK.
Step 6. Set up the added model
Now define the tasks the newly added model will be used for. This is a required step because the ONLYOFFICE AI plugin supports multiple models that can perform certain tasks.
The current version of the AI plugin provides the following options:
- Ask AI lets you interact with the chatbot in a separate window;
- Summarization helps you produce a shorter version of your text;
- Translation allows you to translate your text into/from different languages;
- Text Analysis helps you analyze words and explore their meanings.
Choose Ollama [llama3.2:latest] where necessary and close the AI configuration window.
Based on the steps above, you can add any AI model offered by Ollama and even add a few ones to make the most of artificial intelligence in ONLYOFFICE.
How to use Ollama models in ONLYOFFICE
Now that your AI model is configured and ready to go, you can use it when working on your documents, spreadsheets, presentations and PDF files. To access the AI assistant, you can use the corresponding options on the AI tab or the context menu if you right-click on the selected text with your mouse.
Using Ollama models in ONLYOFFICE Desktop Editors, you can:
- Get your questions answered;
- Interact with the chatbot in a separate window;
- Quickly find information;
- Explore the meaning of words and phrases;
- Generate text content and extract keywords;
- Summarize and translate texts;
- Write code and more.
Interestingly, this is what the Llama 3.2 chatbot says about what it can do:
I can be used in a variety of ways, from helping you plan a vacation to creating art. I’m here to assist you in finding the help or information you need. My strengths include answering questions, generating text and images and even just chatting with you.
Get ONLYOFFICE Desktop Editors
Make document editing more effective with Ollama models integrated into the free ONLYOFFICE office suite for Windows, Linux and macOS:
Have you already tried ONLYOFFICE in combination with local AI models? Share your feedback in the comments below or tweet us at @only_office.
Useful links
AI integration: Together AI / Mistral Al / Groq AI / LocalAI / DeepSeek