Create what you
imagine.
Create with leading open image and video models running on our own GPU infrastructure, then compare and manage every generation from one focused workspace.
Image models
FLUX · Qwen · SD 3.5
Video models
Wan · Hunyuan · LTX-2
Compute
Patchly GPU servers
Your creation
Preview interface
Made with Patchly
One prompt. Endless directions.
From photorealistic scenes to stylized artwork and motion, Patchly is designed to make experimenting with generative media simple.
Infrastructure we operate
Our GPUs. Our models. Your creative workspace.
Patchly deploys and operates open image and video models on our own GPU infrastructure, then makes them available through one simple creative product.
Creators can use powerful open models without provisioning GPUs, maintaining inference runtimes or managing a separate account for every provider.
Our GPU infrastructure
Image and video inference runs on GPU servers deployed and operated by Patchly.
Complete image workflows
Move from text-to-image to variations, image editing, inpainting and model-specific controls.
Native video workflows
Queue text-to-video and image-to-video renders without leaving the same project workspace.
One workspace
The controls creators need, without the infrastructure noise.
Switch models without rebuilding the workflow. Patchly keeps prompts, seeds, aspect ratios, outputs and generation history together so experiments stay reproducible.
Unified model controls
Expose the settings that matter while keeping a consistent creation experience across model families.
Queued generation
Patchly routes work to the right GPU, tracks render progress and keeps long-running video jobs organized.
Reproducible history
Return to the prompt, model and parameters behind an output instead of losing the recipe in a folder.
Open model ecosystem
Start with the models creators already know.
Patchly brings widely used open image and video models into one interface, so creators can choose the right model without changing tools.
Image models
Generate, edit and iterate on still images.
FLUX.1 [dev]
High-quality text-to-image generation and customization.
Qwen-Image
Strong text rendering and precise image editing.
Stable Diffusion 3.5
Flexible quality, speed and fine-tuning options.
Video models
Turn prompts and source images into motion.
Wan 2.2
Open text-to-video and image-to-video generation.
HunyuanVideo 1.5
Efficient, high-quality text and image-to-video.
LTX-2
Open video generation hosted on Patchly infrastructure.
Simple by design
Idea to creation.
Describe the idea
Start with a prompt or source image and choose the output you want to create.
Choose a model
Pick the model and settings that match the result you're after.
Create and compare
Queue generations, review outputs and reuse the settings behind the best result.
Built on Patchly GPU infrastructure
Open models without the setup.
Choose a model and create. Patchly handles model deployments, GPU scheduling, render queues and generated assets behind the scenes.
A few questions.
What is Patchly?
Patchly is a creative AI workspace powered by open image and video models that we deploy and operate on our own GPU infrastructure.
Which models does Patchly support?
Patchly supports FLUX.1, Qwen-Image and Stable Diffusion 3.5 for images, plus Wan 2.2, HunyuanVideo 1.5 and LTX-2 for video.
Who runs the models?
Patchly does. We deploy and operate every model on our own GPU servers, so creators do not need provider accounts, local GPUs or model setup.
Why use one workspace?
Patchly keeps model selection, prompts, parameters, job progress and outputs together so teams can compare results without rebuilding their workflow for every model.