Appendix

December 1, 2025 · View on GitHub

Microsoft Foundry Setup

For this coures you can use GitHub Models or Microsoft Foundry to access LLMs. While GitHub Models is a great option for many users, if you have an Azure subscription, you can use Microsoft Foundry for production-grade AI applications with enterprise features. If you are new to Azure you can get a FREE Azure account with USD$200, by signing up here. This appendix will guide you through setting up a Microsoft Foundry project and deploying models for this course.

Step-by-Step Setup

You can follow the steps below to deploy models in Microsoft Foundry or visit the Deploy an Azure OpenAI model quickstart which will also walk you through the process.

1. Create a Microsoft Foundry Project

  1. Visit the Microsoft Foundry portal
  2. Sign in with your Azure account
  3. Click + New project
  4. Fill in the project details:
    • Project name: langchain-course (or your preferred name)
    • Subscription: Select your Azure subscription
    • Resource group: Create new or select existing
    • Region: Choose a region close to you (e.g., East US, West Europe)
  5. Click Create (the portal will automatically set up the necessary resources)

2. Deploy Required Models

You'll need to deploy two models for this course:

Deploy gpt-5-mini & gpt-5 (Chat Models):

  1. In your project, go to Models + endpoints in the left navigation
  2. Click + Deploy model → Deploy base model
  3. Search for and select gpt-5-mini
  4. Click Confirm
  5. Configure deployment:
    • Deployment name: gpt-5-mini (keep this name for consistency)
    • Model version: Select the latest available
    • Deployment type: Global Standard
    • Click Deploy
  6. Wait for deployment to complete
  7. Follow the same process and deploy gpt-5 as well

Why deploy both models? gpt-5-mini is used throughout the course for most examples (it's faster and more cost-effective). gpt-5 is used in Chapter 1 for model comparison exercises to demonstrate the performance and capability differences between models.

Deploy Text Embedding Model:

  1. Click + Deploy model → Deploy base model again
  2. Search for and select text-embedding-3-small
  3. Click Confirm
  4. Configure deployment:
    • Deployment name: text-embedding-3-small (keep this name)
    • Model version: Select the latest available
    • Deployment type: Global Standard
    • Click Deploy
  5. Wait for deployment to complete

3. Get Your Configuration Values

After deploying your models, you need two pieces of information:

  1. API Key:

    • In your project, go to Overview in the left navigation
    • Find Endpoints and keys
    • Locate your API Key

  2. Endpoint URL:

    • Locate the Azure OpenAI → Azure OpenAI endpoint value (looks like: https://your-resource.openai.azure.com)

4. Add the API Key and Endpoint to Your .env File

Ensure that you add /openai/v1 to the end of your endpoint URL.

# Microsoft Foundry Configuration
AI_API_KEY=your_azure_api_key_here
AI_ENDPOINT=https://your-resource.openai.azure.com/openai/v1
AI_MODEL=gpt-5-mini

Replace your_azure_api_key_here with your actual Azure API key and update the endpoint URL!

Why Microsoft Foundry?

  • ✅ Production-ready: Enterprise-grade infrastructure and SLAs
  • ✅ Higher limits: More requests per minute than free tiers
  • ✅ Additional features: Private endpoints, content filtering, monitoring
  • ✅ Azure integration: Works seamlessly with other Azure services

Back to Course Setup

Once you've completed the Azure setup, return to the main setup guide and continue with testing your setup.

🐛 Troubleshooting

Solutions:

  1. Make sure .env file exists in the project root
  2. Check that .env contains all required variables:
    • AI_API_KEY=your_key
    • AI_ENDPOINT=your_endpoint_url
    • AI_MODEL=gpt-5-mini
  3. No quotes needed around the values
  4. No spaces before or after the =

Issue: "401 Unauthorized" or "Invalid token"

Solutions:

  1. Create a new GitHub Personal Access Token
  2. Make sure you copied the entire token
  3. The token should start with ghp_ or github_pat_
  4. Check for extra spaces in the .env file

Issue: Rate limit errors

Solution: GitHub Models have rate limits. If you hit them:

  • Wait a few minutes
  • The limits reset quickly
  • You can use Microsoft Foundry instead if you went through the optional setup above