Capability · Framework — orchestration

OpenAI SDK (Python)

The OpenAI Python SDK is the de-facto reference for calling hosted LLM APIs. It powers the majority of Python-based LLM apps — directly, via the many OpenAI-compatible providers (Azure, Groq, Together, vLLM, Ollama), and indirectly through frameworks like LangChain that depend on it. It's actively maintained with full typing and async support.

Framework facts

Category: orchestration
Language: Python
License: Apache-2.0
Repository: https://github.com/openai/openai-python

Install

pip install openai
export OPENAI_API_KEY=sk-...

Quickstart

from openai import OpenAI
client = OpenAI()

res = client.chat.completions.create(
    model='gpt-4o-mini',
    messages=[{'role':'user','content':'Summarise the ReAct pattern.'}]
)
print(res.choices[0].message.content)

Alternatives

Anthropic Python SDK — native Claude client
LiteLLM — provider-agnostic wrapper
LangChain / Mirascope — higher-level abstractions
Vercel AI SDK — JavaScript equivalent

Frequently asked questions

Can I use this SDK with non-OpenAI providers?

Yes — any provider with an OpenAI-compatible endpoint (vLLM, Groq, Together, Ollama, Azure OpenAI) works by setting `base_url` and `api_key`. That's why so many tools standardise on this SDK.

Chat Completions or Responses API?

For new code on OpenAI models, prefer the Responses API (it supports hosted tools, built-in retrieval, and MCP). For portability across providers, Chat Completions is still the widest-supported surface.

Sources

OpenAI Python SDK GitHub — accessed 2026-04-20
OpenAI API docs — accessed 2026-04-20