EmpirioLabs AI is a specialized AI inference and integration provider. It self-hosts open-source models on GPU infrastructure it manages, runs optimized endpoints for proprietary models, helps teams ship their own models to large audiences, and offers on-demand GPU Cloud instances and hosted AI agents. Its model catalog covers text, image, audio, video, transcription, embedding, reranking, research, and 3D models.

Nodes

All nodes are in the EmpirioLabs AI category. Each model list loads from the EmpirioLabs AI model catalog when ComfyUI starts and shows only the models that support what the node does.

| Node | Outputs | What it does | | --- | --- | --- | | EmpirioLabs AI Generate Image | images (IMAGE), url (STRING) | Creates images from a prompt. Options: aspect ratio, resolution, number of images (1 to 4) and seed. url links to the first image. | | EmpirioLabs AI Generate Video | video (VIDEO), url (STRING) | Creates a video from a prompt, or animates a start image given as a public link in image_url. Options: duration, resolution, aspect ratio and seed. The node shows the job status and progress while the video renders. | | EmpirioLabs AI Chat | text (STRING) | Sends a prompt, an optional system prompt and optional images to a chat model and returns the reply. Images are sent with the prompt to models that accept image input. | | EmpirioLabs AI Text to Speech | audio (AUDIO) | Turns text into speech. Leave voice empty for the model's default voice. |

Every node also has an optional extra_parameters field for any other request field a model accepts, written as a JSON object, for example {"negative_prompt": "blurry"}. Fields set there take precedence over the node's dropdowns. Each model lists the fields and voices it accepts on its page at empiriolabs.ai/models.

To animate a generated image, connect the url output of Generate Image to the image_url input of Generate Video.

Installation

The pack needs ComfyUI 0.3.48 or later and no extra Python packages.

ComfyUI-Manager. Once the pack is listed in the Comfy Registry, open Manager, choose Custom Nodes Manager, search for EmpirioLabs AI, install it and restart ComfyUI. With comfy-cli, run comfy node install empiriolabs.

Git. Clone the repository into ComfyUI's custom_nodes folder and restart ComfyUI:

cd ComfyUI/custom_nodes
git clone https://github.com/EmpirioLabs-AI/comfyui-empiriolabs.git

API key

Sign in or create an account at platform.empiriolabs.ai.
Open API Keys in the dashboard and create a key. Keys start with sk-empiriolabs-.
Give the key to ComfyUI in one of these two ways.

Environment variable. Set EMPIRIOLABS_API_KEY, then restart ComfyUI:

# Windows (PowerShell or Command Prompt); open a new window afterwards
setx EMPIRIOLABS_API_KEY "sk-empiriolabs-..."

# macOS and Linux; add the line to ~/.zshrc or ~/.bashrc to keep it
export EMPIRIOLABS_API_KEY="sk-empiriolabs-..."

Config file. Copy config.example.ini to config.ini in the pack folder (ComfyUI/custom_nodes/comfyui-empiriolabs, or ComfyUI/custom_nodes/empiriolabs when installed from the registry) and paste the key after api_key =. The nodes read the file on every run, so no restart is needed. Git ignores config.ini, and it is never part of the published pack.

When both are set, the environment variable is used. The nodes never take the key as an input, because ComfyUI saves node inputs in workflow files and in the metadata of the images it saves.

Example workflows

The pack adds two example workflows to ComfyUI's Templates panel, listed under the pack's name. You can also open them from the example_workflows folder.

empiriolabs_generate_image.json: Generate Image, then Save Image.
empiriolabs_image_to_video.json: Generate Image creates the first frame, Generate Video animates it, and Save Video saves the result.

Usage notes

Credits. Requests use the credits on your EmpirioLabs AI account. Prices are listed on the pricing page. Each run with a new seed is a new request; set the seed's control after generate to fixed to reuse the last result while the inputs stay the same.
Waiting. The node shows its status and the elapsed time. Video generation usually takes 1 to 15 minutes.
Cancelling. Cancelling the run in ComfyUI stops the node and cancels the job it is waiting for. A generation that has already started may still be billed.
Generated file links expire. Save the files you want to keep with ComfyUI's Save nodes.
Model lists. The lists refresh in the background every few hours; reload the page to see new models. If the catalog cannot be reached when ComfyUI starts, each node lists one model until the catalog loads. A workflow saved with a model that is not in the current list still runs.
Errors. The node shows the message returned by the API, such as an unsupported parameter for the selected model, with a request ID to quote to support.

Data handling

The nodes send your prompts, inputs and API key only to the EmpirioLabs AI API at https://api.empiriolabs.ai, and download generated files only from the links the API returns. The pack does not store prompts, outputs or keys; ComfyUI saves the files you save with its own nodes. For how EmpirioLabs AI handles data, see the privacy policy.

Development

The tests use a mocked API and never reach the network. Run them from a ComfyUI Python environment and point COMFYUI_PATH at the ComfyUI folder:

COMFYUI_PATH=/path/to/ComfyUI python -m pytest

scripts/live_smoke.py checks the pack against the live API. Without --spend it only reads the public model catalog; each paid step runs one small request:

COMFYUI_PATH=/path/to/ComfyUI EMPIRIOLABS_API_KEY=sk-empiriolabs-... python scripts/live_smoke.py --spend --image --chat --speech

Versions are published to the Comfy Registry by the publish workflow when the version in pyproject.toml changes on main. It needs a Comfy Registry publishing key in the REGISTRY_ACCESS_TOKEN repository secret.

License

MIT

comfyui-empiriolabs

Nodes (4)

EmpirioLabs AI nodes for ComfyUI