Flux Dev (also called FLUX.1 Dev) is a version of Black Forest Labs’ Flux image model.
It’s a high-performance text-to-image AI image generator.
It has open weights. This means you can run it yourself using tools like ComfyUI or the Diffusers library without an API.
How to use it
Head over to OpenArt, pick “Image”, choose “Flux Dev”, and create an image.
Key features
- Generates great images from text, nearly on par with Flux Pro.
- Follows your prompts closely.
- Made efficient via guidance distillation.
- Weights are open — great for research or DIY use.
- Offers a non-commercial license with full image rights.
- Works locally through ComfyUI, Diffusers, or APIs like Replicate and fal.ai.
Best use
Perfect when you want control without paying or relying on cloud services.
Ideal if you’re running things locally, experimenting with prompts, or building your own UI integrations.
My results
I tried this AI image generator.
Here’s the first input I gave it: “A serene morning landscape with a winding path through a foggy forest, soft golden light filtering through tall pines.”
Here’s the result:
But here’s the kicker.
Flux Dev is an AI image generator that allows for advanced guidance. This means you can tweak the inputs to get very accurate outputs.
For example, you can upload a sample image to it so that the AI knows exactly what you’re looking for.
I tried this feature.
Instead of just telling the AI to create the previous image, “A serene morning landscape with a winding path through a foggy forest, soft golden light filtering through tall pines”, I also gave it this real image as guidance:
Here’s what my input looked like:
Here’s the result:
That’s impressive. It’s very similar to the input but clearly different. The only thing I don’t like is those vertical bars across the image.
Let’s try another image with guidance.
I want to create an image like this:
But I want to make it look like “A woman doing dance moves at midnight in New York.”
Here’s how I’m testing it:
But here’s the result:
Yes, it’s clearly another image and looks cool. But the problem is that it doesn’t look like we’re in New York, nor is it midnight.
So this AI image model follows the input images very closely.
I even tried to tweak the creativity level.
- If it’s close to 0, the AI-generated image is essentially the same as the input image.
- But if you put it closer to 1, it gets completely out of hand.
Here’s the same image with a creativity level of 0.93:
But now, this is pretty cool: If you use Flux Dev through OpenArt, there’s a pose editor that looks like this.
Yes, you can literally move the 3D model to any pose you like.
For example, here’s an image I got from the AI with the above pose:
Isn’t that awesome?
This model is customizable, but you have to play with it for a while to get the gist of it.
Here are all the image guidance options:
Make sure to try them all for the best results.
To take home, this model is an advanced AI image generator. If you just tell it to create an image, it’s not that impressive.
But what moves the needle is the guidance. This model follows your prompts very closely and also allows for super advanced guidance, such as the pose or sketch references.