openai-ollama

Create a local Ollama proxy service for the OpenAI compatible backend. This allows you to integrate your OpenAI backend in BYOK mode as an Ollama backend with other applications, such as VSCode GitHub Copilot.

Usage

pnpm i -g openai-ollama

openai-ollama

Or just

pnpm dlx openai-ollama

Configuration

The recommended way is to configure OpenAI compatible backend and server via configuration files.

{
  "baseURL": "<YOUR_BASE_URL>",
  "apiKey": "<YOUR_API_KEY>",
  "models": [
    {
      "id": "<MODEL_ID>",
      "name": "<MODEL_NAME>"
    }
  ]
}

Then just pass the file argument to the CLI:

openai-ollama --config-file=/path/to/config.json

Alternatively, you can configure most options through environment variables, or configure a few options through command line arguments. When used this way, command line arguments have a higher priority than configuration files, while environment variables always have the lowest priority.

Supported options are as follows:

• apiKey: (Required) API key for OpenAI compatible backends. Defaults to OPENAI_API_KEY environment variable.
• baseURL: Base URL for OpenAI compatible backends. Defaults to OPENAI_API_BASE or OPENAI_BASE_URL environment variable, or https://api.openai.com/v1
• models: Specifies the list of available models. The list of models will be obtained through the OpenAI compatible API (/models) if not specified, which is not supported by some backends.
  • Elements in models are objects with the following properties, or the id property of the object.
    • id: Unique ID of the model.
    • name: Display name of the model.
• port or --port: The port the server listens on. Defaults to PORT environment variable or 11434.