Black Forest Labs' Flux.2 line

Run Flux.2 Klein in Your Browser

Klein 9B and 4B, plus the 2,100+ community models built on them. No 9-billion-parameter download, no GPU, no environment to keep alive.

Images generated on Diffus
Behance logoGoogle logoCanva logoSquarespace logoAmazon logo
The difference, not the spec sheet

Why Run Klein Here

The weights are open. What is not free is the disk, the card and the afternoon of setup.

2,100+ Klein Models, Already Hosted

Community checkpoints and LoRAs built on Klein 9B and 4B are on disk and selectable. Paste a Civitai link to add one that isn't.

Browse models

One Node, the Whole Workflow

Generate, edit, inpaint, outpaint and face swap from a single self-contained Klein node in ComfyUI — no graph to assemble before the first image.

Open ComfyUI

Compare It Against the Rest

Flux.1, Flux Kontext, Qwen, SDXL and SD 3.5 are in the same session. Run one prompt through several and keep what wins.

Start generating

Core capabilities

What You Can Do With It

Generation is the first step, not the whole job.

Prompt to picture

Text to Image on Klein

Prompt Klein 9B for quality or 4B when you want the lighter model, at the aspect ratio you need, with the sampler and guidance already tuned for the Flux.2 line.

Selecting a Flux model on Diffus
Edit where you generate

Edit, Inpaint and Outpaint

The Klein node covers editing in the same place you generate — repaint a region, extend the frame, swap a face — without exporting and losing the parameters.

Editing an image on Diffus
Bring your own LoRAs

Stack Community LoRAs

Klein has its own growing LoRA ecosystem on Civitai. Import one by URL and stack it, or fine-tune your own here and use it the moment training ends.

A ComfyUI workflow on Diffus