What it does

RunDiffusion is a platform that puts a long list of image and video generation models in one workspace, so a team does not have to keep a separate subscription for each. The site shows models from providers including Google, OpenAI, Kling, Black Forest Labs, Qwen, ByteDance, Hailuo, Lightricks and Recraft. Beyond plain generation, the workspace covers image editing such as removing objects or replacing backgrounds, style transfer, layered compositions, generative upscaling and training custom models on a brand's own style. Tools can be chained into workflows, and the platform also hosts open source tools including ComfyUI, Auto1111 and Kohya. For teams, the site describes shared workspaces, usage analytics, budgets, model access and spend limits, and SSO. Plugins are listed for Photoshop, Miro, Revit, Rhino and NVIDIA Omniverse, and the examples are grouped by architecture, advertising, fashion, interior design, gaming and design work.

Capabilities

Image and video generation models from many providers in one workspace
Image editing, style transfer, composition and generative upscaling
Custom model training on a brand's own style
Workflows that chain tools together, with shared team workspaces
Open source tools hosted on the platform, including ComfyUI and Auto1111
Plugins for Photoshop, Miro, Revit, Rhino and NVIDIA Omniverse

When to use it

You want to try and compare several image and video models in one place
A design, architecture or advertising team needs shared workspaces with usage analytics and budgets
You want to train a model on your own brand or product look
You want ComfyUI or Auto1111 available inside a hosted workspace