Calling the API

This section shows how to make requests to an LLM provider from Python.

Start here

• Get an API key — pick a provider, sign up, get a key.
• First call — smallest possible "hello world" with the OpenAI SDK.

Going further

• Unified client — target OpenAI, DeepSeek, and Qwen with the same openai client by swapping base_url.
• Streaming — token-by-token output.
• Tool use — let the model call your functions.

One SDK covers three providers

All three providers covered here speak the OpenAI wire format, so a single Python client works for all of them:

flowchart LR
    A[openai Python SDK] --> OA[OpenAI]
    A -->|base_url swap| DS[DeepSeek]
    A -->|base_url swap| QW[Qwen / DashScope]

The practical consequence: learn the openai client well and you can reach every provider in this tutorial.