> ## Documentation Index
> Fetch the complete documentation index at: https://ngrok.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Expose and Secure Your Self-Hosted Ollama API

> Use ngrok to create a public endpoint for your local LLM, enabling remote access and integration, while also adding authentication and access control.

export const domain_0 = "https://your-ollama-llm.ngrok.app"

Ollama is a locally deployed AI model runner, designed to allow users to download and execute large language models (LLMs) locally on your machine.
By combining Ollama with ngrok, you can give your local LLM an endpoint on the internet, enabling remote access and integration with other applications.

But putting your entire Ollama API on the public internet might expose your LLM to abuse—instead, you can use Traffic Policy to add a layer of authentication to restrict access to only yourself or trusted colleagues.

## 1. Reserve a domain

Navigate to the [**Domains** section](https://dashboard.ngrok.com/domains) of the ngrok dashboard and click **New +** to reserve a free static domain like {<code>{domain_0}</code> || `https://your-service.ngrok.app`} or a [custom domain](/gateway/custom-domains/) you already own.

We'll refer to this domain as `$NGROK_DOMAIN` from here on out.

## 2. Create a Traffic Policy file

On the system where Ollama runs, create a file named `ollama.yaml` and paste in the following policy:

```yaml theme={null}
on_http_request:
  - actions:
      - type: add-headers
        config:
          headers:
            host: localhost
```

**What's happening here?** This policy rewrites the `Host` header of every HTTP request to `localhost` so that Ollama accepts the requests.

## 3. Start your Ollama endpoint

On the same system where Ollama runs, start the agent on port `11434`, which is the default for Ollama, and reference the `ollama.yaml` file you just created.
Be sure to also change `$NGROK_URL` to the domain you reserved earlier.

```bash theme={null}
ngrok http 11434 --url https://$NGROK_URL.ngrok.app --traffic-policy-file ollama.yaml
```

## 4. Try out your Ollama endpoint

You can use `curl` in your terminal to send a prompt to your LLM through your `$NGROK_DOMAIN`, replacing `$MODEL` with the [Ollama model](https://ollama.com/library) you pulled.

```bash theme={null}
curl https://$NGROK_DOMAIN/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
        "model": "$MODEL",
        "messages": [
          {
            "role": "system",
            "content": "You are a helpful assistant."
          },
          {
            "role": "user",
            "content": "Hello!"
          }
        ]
      }'
```

## Optional: Protect your Ollama instance with basic auth

You may not want everyone to be able to access your LLM. ngrok can quickly add authentication to your LLM without any changes to your Ollama configuration.

Edit your `ollama.yaml` file and add in the policy below.

```yaml theme={null}
on_http_request:
  - actions:
      - type: basic-auth
        config:
          realm: sample-realm
          credentials:
            - user:password1
            - admin:pasword2
          enforce: true
      - type: add-headers
        config:
          headers:
            host: localhost
```

**What's happening here?**
This policy first checks whether the incoming HTTP request contains the appropriate `Authorization: Basic` header and a [base64-encoded](https://en.wikipedia.org/wiki/Base64) version of one of the `username:password` pairs you specified in `ollama.yaml`.
Only requests with valid Basic Auth are passed through to your ngrok agent and forwarded to your Ollama API.

Restart your ngrok agent to apply the new policy.

```bash theme={null}
ngrok http 11434 --url https://$NGROK_URL.ngrok.app --traffic-policy-file ollama.yaml
```

You can test your policy by sending the same LLM prompt to Ollama's API with the `Authorization: Basic` header, once again replacing `$NGROK_DOMAIN` and `$MODEL`.

```bash theme={null}
curl https://$NGROK_DOMAIN/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H 'Authorization: Basic dXNlcjpwYXNzd29yZDE=' \
  -d '{
        "model": "$MODEL",
        "messages": [
          {
            "role": "system",
            "content": "You are a helpful assistant."
          },
          {
            "role": "user",
            "content": "Hello!"
          }
        ]
      }'
```

If you send the same request *without* the Authorization header, you should receive a `401 Unauthorized` response.

Your personal LLM is now locked down to only accept authenticated users.

## What's next?

* Read more about [Traffic Policy](/traffic-policy), [core concepts](/traffic-policy/how-it-works), and [actions](/traffic-policy/actions) you might want to implement next, like [IP restrictions](/traffic-policy/actions/restrict-ips) instead of Basic Auth.
* Explore other ways to [block unwanted requests](/traffic-policy/examples/block-unwanted-requests/) from Tor users, search or AI bots, and more to protect your self-hosted LLM.
* View your Ollama traffic in [Traffic Inspector](https://dashboard.ngrok.com/traffic-inspector).