Ollama with Open WebUI with support by Elm Computing

Use this path to open the web interface and start chatting with the preloaded model.

1. Launch an EC2 instance from the published AMI.
2. Choose t3.medium or larger for basic validation with the included smollm2:135m model. Larger models require larger instances and more disk capacity.
3. In the security group, allow Open WebUI only from trusted IP addresses.
4. Allow TCP port 8080 from your IP address or trusted network range.
5. Do not expose the Ollama API port 11434 publicly.

After the instance is running, copy the EC2 instance public IPv4 address from the AWS console.

Open this URL in your browser:

URL: http://INSTANCE_PUBLIC_IP:8080

Replace INSTANCE_PUBLIC_IP with the EC2 instance public IPv4 address shown in the AWS console.

On first access, create the initial Open WebUI account. The image includes the small smollm2:135m model so you can validate chat immediately.

If the page does not load, confirm that the instance is running and that the security group allows TCP port 8080 from your current IP address.

Use this section for command-line access, private access through SSH tunnels, direct Ollama API access, or additional model management.

Use the SSH command shown in the EC2 instance console. In the AWS console, select the instance, choose Connect, open the SSH client tab, and copy the generated command.

The command will look similar to this:

Command: ssh -i KEY_PAIR.pem ubuntu@INSTANCE_PUBLIC_IP

Replace KEY_PAIR.pem with your private key file and INSTANCE_PUBLIC_IP with the EC2 instance public IPv4 address.

If you do not want to expose TCP port 8080, use an SSH tunnel from your workstation.

Add this option to the SSH command from the EC2 console:

Option: -L 8080:127.0.0.1:8080

The full command will look similar to this:

Command: ssh -i KEY_PAIR.pem -L 8080:127.0.0.1:8080 ubuntu@INSTANCE_PUBLIC_IP

Then open this local URL from your workstation:

URL: http://127.0.0.1:8080 

If local port 8080 is already in use, choose another local port such as 18080.

Command: ssh -i KEY_PAIR.pem -L 18080:127.0.0.1:8080 ubuntu@INSTANCE_PUBLIC_IP

Then open this local URL from your workstation:

URL: http://127.0.0.1:18080 

Ollama listens locally on the instance at this URL:

URL: http://127.0.0.1:11434 

Keep port 11434 private. To access the Ollama API from your workstation, use an SSH tunnel.

Add this option to the SSH command from the EC2 console:

Option: -L 11434:127.0.0.1:11434

The full command will look similar to this:

Command: ssh -i KEY_PAIR.pem -L 11434:127.0.0.1:11434 ubuntu@INSTANCE_PUBLIC_IP

Then call Ollama locally from your workstation:

Command: curl http://127.0.0.1:11434/api/tags 

From an SSH session on the instance, check the preloaded model.

Command: ollama list

From an SSH session on the instance, run a simple prompt.

Command: ollama run smollm2:135m "Reply with OK only."

From an SSH session on the instance, check the Ollama API.

Command: curl http://127.0.0.1:11434/api/tags 

From an SSH session on the instance, check Open WebUI.

Command: curl -I http://127.0.0.1:8080 

From an SSH session on the instance, use Ollama to pull additional models after launch.

Command: ollama pull llama3.2:1b

Command: ollama list

Choose model sizes that fit the instance memory and disk capacity. The root volume is intended for basic use and validation; increase storage before pulling larger models or maintaining multiple model copies. If model generation is slow, use a larger instance type or pull a smaller model. CPU inference speed depends heavily on instance size and model size.