Skip to main content
Version: 2026 R2

OpenAI

OpenAI offers the GPT family of models — some of the most widely recognized and commonly used language models in the world. They are known for high-quality responses, strong contextual understanding, and a broad ecosystem of documentation, tools, and community support.

When to choose OpenAI​

High-quality responses:

  • tasks that require precise reasoning
  • complex analysis and inference,
  • creative content generation.

Simple configuration:

  • only a single API key is required,
  • no additional infrastructure components are needed, such as a bucket or service account,
  • quick setup that can be completed in just a few minutes.

Proven and mature solution:

  • extensive documentation and many usage examples,
  • broad community support,
  • regularly updated models.

Function calling:

  • strong support for function calling,
  • easy integration with external tools,
  • ability to generate structured JSON responses.

Requirements​

  • an OpenAI account,
  • an OpenAI API key.

Step 1: Get an API key​

  1. Log in to OpenAI Platform.
  2. Go to API Keys.
  3. Click Create new secret key.
  4. Enter a name (e.g., aiproxy-key).
  5. Copy the genereated API key. It typically starts with sk-.
Important

The API key is shown only once, so make sure to store it in a secure location.

Step 2: Configure AI Proxy​

Example aiconfiguration.json​

{
  "ProviderConnections": [
    {
      "Id": "openai-connection",
      "DisplayName": "OpenAI",
      "Type": "OpenAi",
      "ApiUrl": "https://api.openai.com/v1",
      "ApiKey": "sk-your-openai-api-key-here"
    }
  ],
  "ProviderModels": [
    {
      "Name": "gpt-4o-mini",
      "ConnectionId": "openai-connection",
      "ModelName": "gpt-4o-mini"
    },
    {
      "Name": "gpt-4o",
      "ConnectionId": "openai-connection",
      "ModelName": "gpt-4o"
    }
  ],
  "MethodTypesConfiguration": {
    "Chat": {
      "DefaultModel": "gpt-4o-mini"
    },
    "Image": {
      "DefaultModel": "gpt-4o"
    }
  }
}

File docker-compose.yml​

name: aiproxy_containers
services:
  ai-proxy:
    image: webconbps/aiproxy:1.0.0.235
    container_name: ai-proxy
    restart: unless-stopped
    ports:
      - "5298:8080"
      - "7033:8081"
    environment:
    # AiProxy variables
      - ASPNETCORE_ENVIRONMENT=Production
      - AppConfiguration__SelfHosted__Certificate__Path=/app/https/certificate.pem
      - AppConfiguration__SelfHosted__Certificate__Password=

    # Logging
      - Logging__LogLevel__Default=Information
      - Logging__LogLevel__Microsoft=Warning
    volumes:
      - ./certificates/certificate.pem:/app/https/certificate.pem:ro
      - ./aiconfiguration.json:/app/aiconfiguration.json:ro

Troubleshooting​

Error 401 Unauthorized​

  • verify that the correct API key is being used in the configuration,
  • make sure the key is still active.

Error 429 Rate Limit​

  • check the rate limits for your OpenAI account,
  • review the frequency of requests and reduce it if necessary.

Error: Model not found​

  • verify that the model name is spelled correctly,
  • make sure your account has access to the specified model.

Chat and vision models​

  • gpt-4o - a next-generation multimodal model that supports text and images,
  • gpt-4o-mini - a more efficient and lower-cost variant of gpt-4o,
  • o1-preview - a model designed for tasks that require advanced reasoning,
  • o1-mini - a lighter model for tasks that require precise thinking,
  • gpt-4-turbo - a faster GPT-4 variant with a larger context window.
Note

OpenAI models are updated regularly. Information about the latest model versions, such as gpt-4o-2024-11-20, is available in the OpenAI Models documentation.

Embedding models​

  • text-embedding-3-small - a compact model for generating embeddings,
  • text-embedding-3-large - a model designed for higher-quality embeddings,
  • text-embedding-ada-002 - an older embedding model that is still commonly encountered.

Further resources​