FLUX Prompts
Black Forest Labs' image model — it does what the prompt says, renders text you can actually read, and fits any team's stack.
About FLUX
FLUX 2 is the flagship image model from Black Forest Labs, the team behind one of the most widely adopted open-weight image ecosystems. It is available both as a hosted service and as open weights, which is why it shows up everywhere from indie tools to production pipelines.
Its defining strengths are prompt adherence and typography. Where many models treat a prompt as loose inspiration, FLUX follows instructions with unusual literalness — object counts, spatial relationships, layout requests — and it renders legible, well-formed text inside images, which remains a weak spot for much of the field. That combination makes it the pragmatic default for teams that need predictable results.
A model this obedient amplifies whatever you feed it: a precise prompt yields a precise image, and a vague one yields a faithful rendering of vagueness. The gallery below collects tested FLUX prompts with their actual outputs, so you can start from phrasing that is already proven to steer the model.
How to write FLUX prompts
- 1Put any text you want rendered in double quotes and describe its treatment: "SUMMER SALE" in bold condensed sans-serif across the top of the poster. Quoted copy plus placement is the reliable typography recipe.
- 2Keep in-image text short. A few words render crisply; full sentences invite drift. For dense layouts, prioritize the headline and let secondary text stay suggestive.
- 3Write full sentences with explicit spatial language — "a red teapot on the left, a stack of blue cups to its right, window light from behind". FLUX honors these relationships instead of shuffling them.
- 4State counts and attributes exactly: three candles, one lit. Adherence is a strength, so precision in the prompt converts directly into precision in the image.
- 5Commit to one clear style direction per prompt. Stacking contradictory style tags wastes the model's literalness — it will try to honor all of them.
- 6Prompts port well across the FLUX family: iterate on a hosted endpoint, then reuse the same phrasing on open-weight variants in your own pipeline.
Frequently asked questions
What is FLUX 2 best at?
Following instructions precisely and rendering readable text inside images. If your brief involves specific layouts, object counts, spatial arrangements, or visible copy — posters, UI mockups, product shots with labels — FLUX is a dependable first choice.
Can FLUX 2 render text in images?
Yes, and more reliably than most image models. Put the exact wording in quotes, describe the type style and placement, and keep the copy short. The tested prompts on this page show the pattern in action.
Is FLUX 2 open source?
Black Forest Labs maintains an open-weight ecosystem alongside its hosted offering: some FLUX variants ship as open weights you can run on your own infrastructure, while the newest flagship capabilities arrive via the hosted API first. Prompting skills transfer across both.
Where can I try these FLUX prompts?
FLUX is available through Black Forest Labs' FLUX playground and API, and through the many inference platforms and local tools that support its open-weight variants. Copy a prompt from this page and adjust the subject or the quoted text to fit your brief.
Related models
FLUX 3
Black Forest Labs' FLUX 3 — one multimodal flow model for razor-sharp images and 20-second video with native audio.
MiniMax H3
MiniMax H3 is a powerful multimodal video generation model that creates up to 15-second 2K videos with native stereo sound from text, image, video, and audio inputs.
Seedance 2.0
ByteDance's #1-ranked video model — multi-shot direction with native lip-sync and spatial audio.
GPT Image 2
OpenAI's top-ranked image model — it plans the layout, reasons, and self-checks before it draws, nailing complex multi-part instructions.