How to Run Flux Kontext Pro/Max in ComfyUI.. Kind of

Many of you likely have heard or even tried Black Forest Labs latest models, Flux Kontext Pro & Max. If you haven't, Flux Kontext Pro and Max use language based image editing, essentially streamlining your image generation process and bringing the power of autoregressive models like GPT 4o to your PC. Read more about the models below:

https://bfl.ai/models/flux-kontext

However, many of the access points to the Kontext models go through platforms like Fal AI, limiting us to generating inside of these platforms rather than in our Comfy workflows, until now! Sort of.

In our latest video, we show you how to set up Kontext Pro or Max inside of ComfyUI using the native Comfy API nodes and a few business/ad driven examples using Kontext. With this in mind, the generations are not entirely local or free, though image generation is extremely cheap. $5 USD will get you 125 images ($0.04 per image). Additionally, running the node inside of Runpod will require whitelisting. Read more about this in the resources below.

Beyond the generation being paid (though cheap) we currently have found no way to pass generation output from the Kontext API node to any other nodes inside of ComfyUI. This seems to be due to the metadata and color channels of the generated image as far as we can tell, but would love to hear any recommendations to getting past this problem!

Would love to see what you guys have made with Kontext and your thoughts on it + where we're going!


Links:
Patreon: https://www.patreon.com/posts/run-flu...

LinkedIN: https://www.linkedin.com/company/sile...

Chapters:
0:00 - 1:26: Flux Kontext Workflow in ComfyUI
1:27 - 1:47: Cost
1:47 - 6:20: Set Up and Known Issues
6:20 - 8:23: Testing & Results
8:23 - 12:50: Business Use Cases & Limitations
12:50 - 13:34: Overall Summary

#fluxkontext #comfyui