Grok Imagine Image 2.0 API (Grok Imagine Image 2 API) — Python SDK & MCP Server

A focused Python SDK and MCP server for the Grok Imagine Image 2.0 API through MuAPI. Also known as the Grok Imagine Image 2 API or Grok Imagine API, it provides xAI image generation, text-to-image, image-to-image editing, multi-reference generation, local uploads, and asynchronous job polling from Python or an MCP-capable agent.

Availability: The grok-imagine-image-2 endpoint is listed as upcoming in MuAPI's latest model catalog. This client targets the production endpoint contract and is ready to use as soon as access is enabled for your API key.

Related Projects

• MuAPI — Unified API for image, video, and audio generation across hundreds of AI models.

• Grok Imagine Image 2.0 on MuAPI — Official model landing page for Grok Imagine Image 2.0 generation and editing.

• Grok Imagine Image 2.0 playground — Try the model in the browser when access is enabled.

• MuAPI API reference — REST endpoint and asynchronous prediction lifecycle documentation.

• MuAPI access keys — Create the x-api-key credential required by this SDK.

• awesome-ai-image-models — Compare image models by API, price, quality, and use case.

• Awesome-GPT-Image-2-API-Prompts — Reusable prompt patterns for image generation, typography, editing, and visual design.

• Open-Generative-AI — Open-source image and video studio powered by MuAPI.

• Generative-Media-Skills — Agent-ready skills for driving image, video, and audio models from coding assistants.

• muapi-cli — Command-line access to MuAPI image, video, and audio endpoints.

• Wan-3.0-API — the companion Python SDK and MCP server for Wan video generation.

• Flux-3-Dev-API — MuAPI access to FLUX image and video workflows.

Related MCP server: OpenAI GPT-Image MCP Server

Install

git clone https://github.com/Anil-matcha/Grok-Imagine-Image-2-API.git
cd Grok-Imagine-Image-2-API
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env

Set MUAPI_API_KEY in the env file. The client uses https://api.muapi.ai/api/v1 by default. Set GROK_IMAGINE_IMAGE_2_API_BASE_URL to target a compatible self-hosted or proxy endpoint instead. GROK_API_BASE_URL is also accepted as a shorter alias.

Quick start

from grok_imagine_image_2_api import GrokImagineImage2API

api = GrokImagineImage2API()

job = api.text_to_image(
    "A high-contrast halftone portrait in fine white dots on a black background",
    aspect_ratio="1:1",
)

result = api.wait_for_completion(job["request_id"])
print(result)

The API is asynchronous: submit a prompt, keep the returned request ID, and poll until the task is completed.

Image editing and multi-reference generation

Pass one or more public image URLs to edit_image(). The model accepts up to five references in one request, which is useful for combining a subject, location, props, and a target style.

job = api.edit_image(
    prompt="Place the subject in a rainy neon street while preserving their face and clothing.",
    images_list=[
        "https://example.com/subject.jpg",
        "https://example.com/street.jpg",
    ],
    aspect_ratio="9:16",
)

result = api.wait_for_completion(job["request_id"])
print(result)

For a single method that handles both modes, use generate(prompt, images_list=...).

Upload a local reference

uploaded = api.upload_file("reference.png")
print(uploaded)

Use the URL returned by the upload endpoint in images_list for a later generation or edit request.

API surface

Supported aspect ratios

The current catalog contract supports:

1:1, 1:2, 2:1, 9:16, 16:9, 2:3, 3:2, 3:4, and 4:3.

MCP server

Expose the model to MCP-capable clients:

python mcp_server.py

The server provides text_to_image, edit_image, generate_image, and get_task_status tools. Configure it in an MCP client with the repository's Python interpreter and pass MUAPI_API_KEY through the process environment.

Example configuration:

{
  "mcpServers": {
    "grok-imagine-image-2": {
      "command": "/absolute/path/to/.venv/bin/python",
      "args": ["/absolute/path/to/Grok-Imagine-Image-2-API/mcp_server.py"],
      "env": {
        "MUAPI_API_KEY": "your_muapi_api_key"
      }
    }
  }
}

Endpoint compatibility

The client calls these MuAPI paths beneath the configured base URL:

• POST /grok-imagine-image-2

• POST /upload_file

• GET /predictions/{request_id}/result

The SDK uses the x-api-key header and JSON request bodies. The model endpoint accepts prompt, optional images_list, and aspect_ratio.

Development

Run the local tests and syntax checks with:

python -m unittest discover -s tests -v
python -m py_compile grok_imagine_image_2_api.py mcp_server.py

License

MIT