Flux

basicwidespread

Flux is the image-generation model family of the German Black Forest Labs, which creates realistic images from text descriptions and can also edit existing images.

In plain terms

Flux is like a very fast illustrator: you describe in words what you want to see, and it draws it within seconds. You can request photorealistic images, drawings, posters, or product photos.

It became known mainly for following longer descriptions precisely, writing requested image captions mostly legibly, and making fewer errors with hands and faces than previous models.

It is not a single program but a model family. There is a fast, free-to-use version, one for developers, and a paid, top-quality version. Some versions can be downloaded and run on your own machine, which is rare among major competitors.

You often use it without realizing: other companies' image editors, chatbots, and design tools integrate it in the background.

How it works

The model was trained on countless images and their corresponding descriptions, so it learned which forms, colors, and compositions correspond to which words.
When you enter your request, a text-understanding part converts it into numbers so the image-generation part can "understand" what to depict.
The image starts from random visual noise, like a snowy TV screen. The model shapes this into the described image over multiple steps.
Flux uses a method called flow matching for this: it learns the straightest possible path from noise to the finished image, so fewer steps are needed. The fastest versions were specifically trained to produce good images in just a few steps.
For editing variants (such as Flux Kontext), you can provide an existing image and ask in text what to change while the rest stays unchanged.

An everyday example

Your small business webshop needs atmospheric product photos, but you have no money for a photographer. In an online image editor, you type 'ceramic mug on a wooden table, morning light, coffee beans next to it', and Flux in the background creates several variations in half a minute for you to choose from.

Who is affected

marketerYou can create campaign images, sketches, and product visualizations in minutes without photography or purchasing stock photos. → Marketing

entrepreneur, executiveIt makes visual content production cheaper and faster, but you must check the license of the specific version before commercial use.

developerOpen-weight versions can be integrated into your own application, fine-tuned, or invoked as an agent tool. → Code

everyoneAn increasing share of images seen online are created with such models, so it is worth knowing that even a lifelike photo is not necessarily real.

A common misconception

MythFlux is completely free and open-source, usable freely for anything.

In realityOnly certain versions are freely usable. Developer versions have licenses that restrict commercial use, while the most powerful versions are available as paid services.

Limits and risks

It can also produce deceptively lifelike fake images of real people or events, so you should check the source of any image critically.
The copyright assessment of images used for training and the legal protection of generated images are disputed in several countries.
Running it on your own machine requires a powerful graphics card; on an average laptop, it is slow or will not start at all.
It can still make mistakes: it may ruin longer texts, miss small details, or ignore part of your description.

Similar, but not the same

Stable Diffusion – Stable Diffusion is a model family associated with Stability AI, launched earlier. Flux was partly created by the same researchers who built the foundations of Stable Diffusion, but now at a new company, Black Forest Labs.
Midjourney – Midjourney is a closed, subscription-based service, while certain versions of Flux are downloadable and can be integrated into other products.
Flux (software development tools) – The same name is used by other IT tools (for example, a deployment automation tool for Kubernetes systems), which have nothing to do with image generation.

Related concepts

Broader

Generative AI, Image generation, text-to-image model

Narrower

Flux Kontext, Flux Schnell, Flux Dev, Flux Pro

On the edge

Stable Diffusion, Midjourney, Diffusion model, Deepfake, Prompt

A brief history

Black Forest Labs was founded in 2024 by key figures behind the Stable Diffusion research, including Robin Rombach. The first model family, FLUX.1, was released in August 2024; the image-editing FLUX.1 Kontext followed in 2025, and FLUX.2 arrived at the end of 2025.

Where things stand

At the end of 2026, Flux is one of the best-known image generation model families, offered built in by numerous image editors, design programs and cloud services. A significant community has formed around the freely downloadable versions, creating its own fine-tunings and add-ons.

Status: 2026 November

In the news

Appears in 1 Agent Age stories; the curve shows weekly occurrence in the news magazine.

Flux expands platform to show whether AI coding investments are paying offSep 23, 2026 · Code

Also known as

Flux · FLUX · FLUX.1 · FLUX.2 · Flux Kontext · Flux AI · Black Forest Labs Flux · BFL Flux

Sources

This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.