Ollama + WebUI

Ollama with Open WebUI is a ready-to-use solution for local and cloud deployment of modern language models with a convenient web interface. The product includes fully configured Docker containers:

• Ollama — a platform for downloading, running, and managing open-source LLMs on your own hardware without dependency on cloud services.
• Open WebUI — an extensible and convenient web interface for working with models, supporting Ollama and compatible APIs.

When installing the product, you need to specify the model that will be preloaded, for example deepseek-r1:8b. Additionally, you can download any other models from the Ollama catalog: both during product installation and during use.

Solution advantages:

• Simple deployment through Marketplace.
• Support for various LLMs in one interface.
• Offline operation and full control over data.
• Intuitive web interface for interacting with models.

1. Get an SSH key pair to connect to the virtual machine (VM).

2. Create a service account without a role.

3. Create a VM from a public image. In the Image/boot disk selection block, go to the Cloud Marketplace tab and select Ollama + WebUI. In the Access block:

   • in the Login field, enter the username;

   • in the SSH key field, paste the contents of the public SSH key file.

4. Make sure that in the selected Security group the incoming port TCP/22 is allowed.

5. In the Product configuration form, specify the model name from the Ollama catalog, for example, deepseek-r1:8b

6. In the Additional block, specify the previously created service account.

7. Click “Create VM” and wait for the deployment process to complete.

8. To connect to the web interface, you need to create an SSH tunnel:

   ssh -L 8080:localhost:8080 <username>@<VM_public_IP_address>
   

9. The Ollama WebUI web interface will be available at http://localhost:8080 after loading the model selected during installation (for the deepseek-r1:8b model this may take several minutes).

• Text generation — from stories and articles to posts and descriptions.

• Summarization — brief summary of long documents, articles or reports.

• Data analysis — extracting key insights and identifying patterns.

• Programming assistance — code examples, algorithm explanations, error correction.

• Translation and language work — text translation, style adaptation, improving formulations.

• Productivity assistant — preparing notes, plans, ideas and drafts.

Yandex Cloud technical support responds to requests 24 hours a day, 7 days a week. Available types of requests and their processing time depend on the pricing plan. You can connect paid support in the management console. Learn more about the procedure for providing technical support.

Yandex Cloud does not provide technical support for the product. If you have problems, contact the developer’s information resources.