First of all you need to log in to your profile or register on the Cabina.AI website. Next, for getting started with Flux AI it is enough to describe the picture you want to get in a special input line, as if you were explaining your idea to an artist. For example: “Orange sunset over the sea with waves, a lonely palm tree stands on the shore, a flock of birds flies in the sky.” The more detailed the description, the more accurate the result.

After registration, you get free tokens that you can spend on the Flux online image generator. You can always top up your balance with one of the suggested amounts or purchase a monthly subscription that suits your needs. More details on the price page

FluxAI can create almost any image you want, the only limit is your imagination. Experiment with different prompts, create multiple images at once, choose the orientation of the image. All these little tips will allow you to create a beautiful result. Important: the clearer the prompt, the better the image you will get. 

Flux AI, like other AI image generators, is primarily used to create images from text descriptions. It is ideal for: creating original art images, banners, images for Instagram, Facebook, creating images for blog posts, creating characters for video games, etc.

Flux is not an LLM. It's a text-to-image diffusion model family from Black Forest Labs. “LLM” refers to large language models like GPT-5 or Claude, which generate text. FLUX.2 generates images from text prompts and optional reference images. Cabina.AI hosts both categories side-by-side, which is probably where the naming confusion comes from - but technically, “flux llm model” is a misnomer.

Same underlying models - Max, Pro, Flex, Klein - but billed from one shared token balance instead of a standalone subscription, inside the same interface as every other model Cabina.AI supports.

Yes. FLUX.2 was specifically trained for production-grade typography. UI labels, product copy, and multi-line editorial text render legibly. It's not perfect on every font at every size, but it's the current state of the art for text-in-image.

Yes - specify hex codes directly in the prompt and FLUX.2 matches them closely rather than approximating.

Pro and Flex typically return 4 MP images in under 10 seconds. Klein low-resolution draft mode is near-instant, which is what makes rapid iteration cheap.

Supply up to 10 reference images in a single prompt. FLUX.2 multi-reference conditioning is built for character consistency, product placement, and style transfer across a series.

Monthly plans charge automatically each month with tokens credited instantly. Annual plans charge once a year with the full annual token amount credited immediately. Unused tokens after a subscription ends are preserved for at least 1 year.