Image Generation

Generate images from text prompts using diffusion models, served on the diffusers/generative backend.

Active public models (text-to-image)

Model	Endpoint	Source repo	Notes
Stable Diffusion XL	hf-stabilityai-stable-diffusion-xl-base-1.0	stabilityai/stable-diffusion-xl-base-1.0	High-resolution, ready
Stable Diffusion 3 Medium	hf-stabilityai-stable-diffusion-3-medium-diffusers	stabilityai/stable-diffusion-3-medium-diffusers	MMDiT architecture
FLUX.1-dev	hf-black-forest-labs-FLUX.1-dev	black-forest-labs/FLUX.1-dev	High quality, Black Forest Labs
Z-Image-Turbo	z-image-turbo	Tongyi-MAI/Z-Image-Turbo	Fast, few-step

All active endpoints are currently free. Several are candidate (functional, not fully validated). Verify readiness in the live catalog:

curl "https://api.colabhive.com/api/builder/v1/endpoints?visibility=public&task_type=text-to-image"

You can import any HuggingFace image model with task_type="text-to-image".

Quick start

from colabhive import ColabHive

client = ColabHive(api_key="hive_...", account_id="YOUR_ACCOUNT_ID")

result = client.endpoints.infer(
    endpoint_id="hf-stabilityai-stable-diffusion-xl-base-1.0",   # SDK resolves the name to a UUID
    input_data={
        "prompt": "A sunset over mountains, oil painting style",
        "negative_prompt": "blurry, low quality",
        "width": 1024,
        "height": 1024,
        "num_inference_steps": 30,
        "guidance_scale": 7.5,
    },
)

for art in result.get("result", {}).get("output_artifacts", []):
    client.endpoints.download_artifact(art["url"], f"./{art['filename']}")
    print(f"Saved {art['filename']} ({art['size_bytes']} bytes)")

Generation can take a while and may cold-start; poll GET /tasks/{task_id} if the call returns {"status": "queued"}.

Input parameters (text-to-image)

Parameter	Type	Default	Description
prompt	string	required	Text description of the image
negative_prompt	string	""	What to avoid
width	integer	512	Image width
height	integer	512	Image height
num_inference_steps	integer	30	Quality vs. speed
guidance_scale	number	7.5	Prompt adherence strength
num_images	integer	1	Images to generate
seed	integer	random	For reproducibility

Output format

{
  "output_artifacts": [
    {
      "filename": "image_0.png",
      "url": "https://storage.colabhive.com/...",
      "content_type": "image/png",
      "size_bytes": 1048576,
      "modality": "image"
    }
  ],
  "metadata": { "width": 1024, "height": 1024, "steps": 30 }
}

Fine-tuning (LoRA / DreamBooth)

See the Generative Models Guide for fine-tuning image models on your own data.

Related

• Generative Models · Audio · Video
• Inference API

---

Authors: José Luis Minich, Maximiliano Lucius.