Documentation · Getting started · Models · freetheai.org

FreeTheAI Python SDK

The official Python client for FreeTheAI, a free OpenAI-compatible AI API. One key gives you dozens of chat models, streaming, tool calls, web search with sourced answers, and image and video tasks.

pip install freetheai

Quickstart

Create a key at freetheai.org, do the daily check-in for free models, and set FREETHEAI_API_KEY.

from freetheai import FreeTheAI

client = FreeTheAI()

reply = client.chat.completions.create(
    model="fta/zai/glm-5.3",
    messages=[{"role": "user", "content": "Write a haiku about open models."}],
)
print(reply["choices"][0]["message"]["content"])

Stream it:

for text in client.chat.completions.stream_text(model="fta/kimi/k3", messages=messages):
    print(text, end="", flush=True)

Search the web:

answer = client.answer("What changed in Python 3.14?")
print(answer["answer"], answer["sources"])

Built for the FreeTheAI API

Generic OpenAI clients treat every error the same. This one knows how the FreeTheAI API answers:

• Busy models wait for the server's Retry-After, or move straight to your fallback_models.
• Concurrency limits are learned on the first refusal; extra calls queue instead of failing.
• Long replies are fetched as a stream and assembled, so they don't hit idle timeouts.
• Cut-off streams raise StreamInterruptedError with the partial text, never a silently short answer.
• The daily check-in and daily limit raise CheckinRequiredError and DailyLimitError at once, with the link or reset time.
• Mistyped model IDs come back with suggestions; a missing fta/ prefix is fixed for you.

client = FreeTheAI(fallback_models=["fta/zai/glm-5.3-flash", "fta/mmx/minimax-m3"])

Read Reliability for the full list.

Documentation

More from FreeTheAI

• JavaScript / TypeScript SDK and Go SDK
• Cookbook: runnable examples
• Setup guides for SillyTavern, JanitorAI, Claude Code, Cline, and more
• API status and Discord

License

MIT