FLUX 3

FLUX 3 is an AI model that turns written descriptions into pictures or videos. Just as you might tell a painter, "a sunset beach with a single paper boat," and get a painting, you type a sentence into FLUX 3 and it produces an image or a short video.

The video version, released in August 2026, can generate 20-second clips complete with sound. However, this version reportedly launched without the ability to understand Korean-language prompts.

FLUX 3 was developed by a company called Black Forest Labs, and other companies, such as ElevenLabs, embed the model inside their own services. In other words, the company that built FLUX 3 is different from the companies that license it and offer it inside their own products.

Articles describe things like "ElevenLabs added Black Forest Labs' FLUX 3 and ByteDance's Seedance 2.5 to its video creation tool." It's important not to confuse this: FLUX 3 wasn't built by ElevenLabs — it's an outside model that ElevenLabs offers as an option inside its own tool.

You can test FLUX 3-based models on image or video generation services that support them by entering a prompt like:

A rainy night in a city alley, neon signs reflected on the wet road, the back of a person holding an umbrella, cinematic lighting

The more specific you are about setting, lighting, and mood as you vary the prompt, the closer the result will match what you intended.