Quick SummaryThrough Xano’s Microservice feature, you can deploy Ollama as a part of your Xano instance, enabling secure communication with an off-the-shelf or customizable LLM tailored specifically to your needs.

​
What is Ollama?

Ollama is a platform that enables seamless integration of large language models (LLMs) into various applications. It provides an environment where businesses can deploy pre-built or customized LLMs, facilitating secure and efficient communication tailored to specific business needs. This makes it easier for companies to leverage advanced AI technology without extensive in-house development, and without the concerns of sending data to a third-party AI provider. In addition, deploying Ollama as a part of your Xano instance can be a significant cost-saving measure in comparison to leveraging a third party service.

​
What can I do with Ollama + Xano?

Deploying Ollama as a microservice in Xano helps address a critical data privacy challenge when building AI-enabled applications that handle sensitive information. By processing data within your controlled infrastructure boundary rather than sending it to external AI providers, using Xano with microservices removes one significant barrier to developing secure applications that can still leverage the benefits of AI models.
While this approach addresses an important technical aspect of data privacy, comprehensive security and compliance will require implementing additional safeguards appropriate to your specific regulatory environment.

​
Getting Started

1

If you're not on an Enterprise plan, reach out to our team to get started.

Talk to Sales
2

Head to your instance settings, and select Microservices

3

Choose a model to deploy

You’ll want to know which model you’re deploying so we understand the resources that need to be allocated. Check out the resource linked below if you need help choosing a model.Choosing a Model
4

Add a persistent volume

A persistent volume is just a place that the microservice can store data that remains between restarts. Ollama will use this volume to store the model(s) that you’re working with.
How much storage do I need?When browsing Ollama models, use the Size column for your chosen model to determine how large of a volume you should deploy. Make sure to add a little extra when creating your volume, just for some breathing room.
Name the volume ollama, select the size, and choose SSD as the storage class. When you’re ready, click Add
5

Add a deployment

Click Add under DeploymentsFill out the following information. For this example, we’ll be deploying the phi3:mini model, which should work with the following example values.
ParameterPurposeExample Value
Deployment NameName of the deploymentollama
ReplicasNumber of container instances1
Docker ConfigSource of the Docker imagepublic repo
Deployment StrategyUpdate strategyRollingUpdate
Container NameName of the containerollama
Container TypeContainer typeStandard
Docker ImageDocker image to useollama/ollama
Container PortPort the container listens on11434
Service PortPort exposed to the service11434
Persistent Volume NameName of the persistent storage volume you created in the previous stepollama
Volume TypeVolume typePersistent Volume
Mount PathPath in container where volume is mounted/root/.ollama/models
Min CPUMinimum CPU allocation500m
Max CPUMaximum CPU allocation2000m
Min RAMMinimum RAM allocation4096Mi
Max RAMMaximum RAM allocation8192Mi
Click Add and then Update & Deploy
6

Deploy your chosen model

All interactions with Xano microservices are facilitated by the Microservice function, which is a REST API request to your chosen microservice.Use the cURL below to get started. Add a Microservice function to your function stack, and click Import cURL
curl -X POST http://localhost:11434/api/pull \
  -H "Content-Type: application/json" \
  -d '{
    "name": "phi3:mini"
}'
Please note that the Content-Type: application/json header is required, or your requests will fail.
Click on the host dropdown and choose your deployment, likely called ollamaChange the timeout to a value that will allow you to monitor the deployment right inside of Xano. Below are some recommended values. You can also monitor the deployment via the logs of the microservice, available from your instance settings where you deployed the microservice earlier.
phi3:mini (~5 GB)10–60 seconds5 min (300s)
mistral:7b (~8 GB)30–120 seconds5–10 min (300–600s)
llama3:70b (~40 GB)5–20 minutes or more15–30 min
Once the model has downloaded, you’ll be returned a 200 status response, as shown below. The model will be saved to your persistent storage volume, so you should only have to do this once.
7

Interact with your Ollama deployment

You’re ready to receive generations from your Ollama microservice. Here’s an example cURL command to get you started.
curl -X POST http://localhost:11434/api/generate \
  -H "Content-Type: application/json" \
  -d '{
    "model": "phi3:mini",
    "prompt": "Explain the difference between supervised and unsupervised learning.",
    "stream": false
}'
The above command will return an output like the one shown below.
8

What's next?

Ollama offers a set of standard commands that can be issued via REST API endpoints. You can review them here, and use them as necessary inside of your function stacks.