Civitai Python Client — Retired

Warning

This project is no longer maintained and has been retired.

civitai-py was a hand-written Python client for Civitai's image-generation (orchestration) API. It has not been updated since June 2024, does not track the current API, and will receive no further fixes or releases. The civitai-py package on PyPI is frozen at its last version — do not use it for new projects.

Use the official API

Civitai's APIs are now fully documented at developer.civitai.com. They are plain HTTPS + JSON and work from any language, so there's no need for a bespoke SDK. Point your code — or your coding agent — at the docs and call the API directly.

You want to…	Go here
Generate images, video, audio, and more (what this client did)	Orchestration API guide
Browse & search models, images, creators, tags	Site API guide
Ready-made request examples per model	Recipes
Connect an AI agent / assistant	MCP server
Sign in users with their Civitai account	OAuth guide

Prefer a typed client? Generate one from the OpenAPI spec

The orchestration API publishes a live OpenAPI 3.1 spec that stays in sync with the API on every deploy. Instead of depending on a stale hand-written SDK, generate a fresh, fully-typed client for your language directly from it:

https://orchestration.civitai.com/openapi/v2-consumers.json

Python (openapi-python-client):

pipx run openapi-python-client generate \
  --url https://orchestration.civitai.com/openapi/v2-consumers.json

Any language (openapi-generator):

openapi-generator-cli generate \
  -i https://orchestration.civitai.com/openapi/v2-consumers.json \
  -g python \
  -o ./civitai-orchestration-client

The generated client is always current, unlike this repo. A browsable reference built from the same spec lives at developer.civitai.com/orchestration/reference.

Quick start (raw HTTP)

1. Grab an API key from your Civitai account settings.
2. Submit a workflow to the orchestration API with a bearer token:

curl -X POST "https://orchestration.civitai.com/v2/consumer/workflows?wait=60" \
  -H "Authorization: Bearer $CIVITAI_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "steps": [{
      "$type": "imageGen",
      "input": {
        "engine": "sdcpp",
        "ecosystem": "sd1",
        "operation": "createImage",
        "model": "urn:air:sd1:checkpoint:civitai:4384@128713",
        "prompt": "masterpiece, best quality, portrait, cinematic lighting",
        "negativePrompt": "worst quality, low quality, blurry",
        "width": 512,
        "height": 512,
        "steps": 20,
        "cfgScale": 7
      }
    }]
  }'

See the recipes for per-model request bodies (Flux, SDXL, WAN video, upscalers, TTS, transcription) and authentication for token handling.

Building an AI agent?

Point it at the Civitai MCP server rather than writing a client by hand. There's also a machine-readable index of the docs at developer.civitai.com/llms.txt (and llms-full.txt) for feeding into LLM tools.

---

The original SDK source remains in this repository's history for reference only. It is unsupported.