Flux Pro

This is the core of the model. A detailed prompt yields lighting, skin texture, and material surfaces that read as photographic rather than painted. Black Forest Labs positions FLUX1.1 [pro] as state-of-the-art on prompt following, visual quality, image detail, and output diversity at the same time, a combination earlier models struggled to hold together.

Most image models mangle words. Flux Pro keeps them sharp, which is why teams reach for it when a poster, package, or storefront sign has to stay legible. The model trained on a mix of natural and synthetic data so it can handle long, clean strings of text, not just a stray word or two.

The 1.1 release runs about six times faster than the original FLUX.1 [pro] generation, per Black Forest Labs, with lower latency to boot. Fast inference matters when a campaign needs dozens of variations, since a slow model turns iteration into a waiting game.

FLUX1.1 [pro] Ultra raises output to roughly 4 megapixels, enough for print and for images headed into post-production. It supports common aspect ratios such as 1:1, 16:9, 9:16, 4:3, and 3:4, so the same prompt can feed a square feed post or a wide banner.

Raw mode returns a less processed, more natural-looking image with a softer finish. It suits editorial work and shots that a retoucher will polish later, where an over-styled default would get in the way.

The BFL API exposes model choice, image resolution, and content moderation as settings, so teams can tune output to their own standards. The same models also run on fal.ai, Replicate, Together AI, and Freepik, which gives you a fallback if one host is down.

The API accepts batch requests, so a single job can return many images from a list of prompts. That cuts the overhead of firing off one request at a time when a project needs volume.