FLUX 3 Image: New AI tool creates sharp 4K images

Black Forest Labs, a German AI company founded by former researchers from Stability AI, has released FLUX 3 Image. The tool is part of the larger FLUX 3 model, which the company presented in July and which can also create video and audio. FLUX 3 Image was made available on October 1.

The new tool can create images in up to 4K resolution, a very high level of detail used for things like print and large screens. By comparison, many AI images today are made only at lower resolution, which looks blurrier when enlarged.

Moves and fixes individual elements

One of the most talked-about features is that users can determine exactly where a specific subject should be placed in the image, instead of just describing it in a text prompt and leaving the rest to guesswork. This is called subject positioning.

At the same time, you can do what the company calls partial editing: You fix only a specific area of an image, such as a person's jacket or the background, while the rest of the image stays unchanged. This makes it easier to refine an image over several rounds without having to generate the whole thing from scratch. The tool also supports up to ten reference images in the same generation, so you can combine multiple subjects.

FLUX 3 Image is for now only available through Black Forest Labs' paid API, meaning access that companies and developers can build on in their own products. However, the company has announced that an open version, called FLUX 3 Dev, is coming within the next few weeks. An open version means the model's architecture will be made freely available to download and use, so more developers and smaller companies can build on it without paying for access through the company's own services. No exact date has been set yet, and the company has not said under what terms the model will be released.

What it means for you

If you work with marketing, design or other content, FLUX 3 Image can give you sharper images that also hold up in print and on large screens. The ability to fix individual parts of an image makes it easier to refine results without starting over. Right now it requires access through the company's paid API, but if you'd rather wait a couple of weeks, you can try the free, open version instead.