Set up Ollama, NGROK, and LangChain

Source: Dev.to

This post shows how to quickly set up Ollama with ngrok and use it in LangChain.

Ollama

You can easily download Ollama for Linux, macOS, and Windows:

👉

Test if it’s running

curl http://localhost:11434; echo   # should output “Ollama is running”

Note: If nothing is returned, start the server with ollama serve.

Pull your preferred models

ollama pull phi4-mini
ollama pull nomic-embed-text:v1.5

Ngrok

Download ngrok for Linux, macOS, and Windows:

👉

Note: Remember to add your authtoken!

ngrok config add-authtoken 

Expose Ollama with ngrok (basic auth)

ngrok http 11434 \
  --host-header="localhost:11434" \
  --basic-auth="username:password"

The command generates a public URL, e.g.:

https://09c6b3946ddd.ngrok-free.app

You can test it in a browser:

After successful authentication you’ll see “Ollama is running”:

You can also verify the models via the /api/tags endpoint:

✨ The two models pulled earlier are now ready to use.

LangChain & Ollama

Install the required Python packages:

pip install langchain langchain-core langchain-ollama

Use Ollama (LLM & Embeddings) from another network

import base64
from langchain_ollama.llms import OllamaLLM
from langchain_ollama.embeddings import OllamaEmbeddings

BASE_URL = "https://09c6b3946ddd.ngrok-free.app"
user_pass = b"username:password"
auth_header = base64.b64encode(user_pass).decode("utf-8")

client_kwargs = {
    "headers": {
        "Authorization": f"Basic {auth_header}"
    }
}

llm = OllamaLLM(
    model="phi4-mini:latest",
    base_url=BASE_URL,
    client_kwargs=client_kwargs,
)

embeddings = OllamaEmbeddings(
    model="nomic-embed-text:v1.5",
    base_url=BASE_URL,
    client_kwargs=client_kwargs,
)

# Test the LLM
print(llm.invoke("Hello, Ollama LLM =)"))

# Test the embeddings
print(embeddings.embed_query("Hello, Ollama Embeddings =)"))

Result example:

More Info

• Expose and Secure Your Self-Hosted Ollama API