Model · Black Forest Labs
FLUX 3, image and video that follow the prompt
FLUX comes from Black Forest Labs, the team behind Stable Diffusion. Its reputation rests on one thing: it does what you asked. With FLUX 3, that adherence moves to video: text, an image, a first and last frame, or up to ten keyframes become a 5 to 20 second shot.
What FLUX does better
Prompt adherence
This is its signature. When a brief names nine objects, their place in a grid and the direction of the light, most models drop three of them or rearrange the composition; FLUX follows the brief.
Video, draft first
FLUX 3 can produce a cheaper draft to validate a movement before committing. The full-quality render can then pick up that draft, when the provider returns its cache.
Keyframes to direct the shot
Up to ten keyframes placed in time, or a first and a last frame: the shot goes through the steps you chose. An existing video can also be extended or changed from a brief.
Editing without disturbing the rest
FLUX 3 Image edits from one to ten input images while leaving the rest of the frame alone — the behaviour you expect from a retouch, and not always the one you get.
From 21:9 to 9:16, up to 4K
FLUX 3 Image offers eight formats, from 21:9 to 9:16, at 1K, 2K or 4K. Video covers seven formats, from 21:9 to 9:16, at 720p or 1080p.
The FLUX engines available here
FLUX 3 for image and video, FLUX 3 Max for video, FLUX.3 Klein and Klein Base for images, and FLUX.2 Pro, still in the picker.
- FLUX 3 Image — text to imageEight formats from 21:9 to 9:16, at 1K, 2K or 4K.
- FLUX 3 Image — image editingOne to ten input images and a change described in plain language.
- FLUX.3 Klein — text to image and editingText to image, or editing from one to ten reference images; 1K, 2K or 4K, eight formats from 21:9 to 9:16.
- FLUX.3 Klein BaseThe Base variant of Klein: same inputs, same resolutions and same formats, for text to image and editing alike.
- FLUX 3 — videoText, image, first and last frame, or up to ten keyframes to video. 5 to 20 s, 720p or 1080p, with a draft mode.
- FLUX 3 — extend or edit a videoExtend an existing video, or change it from a brief (720p, source under 15 seconds).
- FLUX 3 Max — videoA post-trained variant of FLUX 3 video: text, a start image (and last frame) or up to ten keyframes to video. 5 to 20 s, 720p or 1080p, native audio; no draft mode, extension or video editing.
- FLUX.2 Pro — text to imageGeneration from the prompt alone, nine ratios, jpeg or png output.
- FLUX.2 Pro — image editingTakes up to nine input images and applies a change described in plain language.
Why run it here
Prompt adherence pays off most when the prompt is generated, not hand-written.
- Feed FLUX from a text node: a list of products becomes a consistent series of images.
- Validate a movement as a FLUX 3 draft before running the final version.
- Compare FLUX, Nano Banana and Seedream on the same brief before deciding.
- Chain a FLUX 3 image into FLUX 3 video in the same graph.
- Ship the workflow as an app so others can rerun it with their own inputs.
- One credit balance, no API keys to manage.
Frequently asked questions
- Which version of FLUX is available?
- FLUX 3, for image (text-to-image and editing) and for video, FLUX 3 Max for video, FLUX.3 Klein and Klein Base for images, and FLUX.2 Pro for text-to-image and editing. The model picker always shows what is actually executable at any given moment.
- Does FLUX 3 do video?
- Yes. FLUX 3 turns text, an image, a first and last frame, or up to ten keyframes into a 5 to 20 second video, at 720p or 1080p. It can also extend or change an existing video.
- What is the FLUX 3 draft mode for?
- To validate a movement at a lower cost before committing. The full-quality render can then pick up the draft, when the provider returns its cache.
- Can FLUX edit an existing image?
- Yes. FLUX 3 Image and FLUX.3 Klein take one to ten input images and FLUX.2 Pro up to nine: you describe the change in plain language, with no mask to trace.
- Why pick FLUX over another model in the catalogue?
- For prompt adherence. When a brief lists specific elements and their layout, FLUX respects them better than average. For text inside the image, Nano Banana remains the better choice; for photographic material, Seedream.
- Do I need a Black Forest Labs API key?
- No. Beemm Vision calls the provider for you and bills in credits on a single balance, shared with every other model in the catalogue.
Try FLUX
Write a detailed brief and see what it does with it. The full catalogue is available from the free plan.
Keep reading
- Edit a photo with AI, by painting or by describingErase an object, replace an area, extend the frame, upscale or remove the background: AI photo editing in 4 steps, cost shown before you run.
- Best AI video generator: the comparisonKling, Seedance, Wan, MiniMax H3, Grok, FLUX 3 and Cinema Studio side by side: length, 4K, sound, cost.
- Recraft on Beemm VisionRecraft V4.1 Pro generates images and, above all, VECTOR output: logos, icons, pictograms. The only model in the catalogue that returns editable curves.
- GPT Image on Beemm VisionUse OpenAI's GPT Image 2 and GPT Image 1.5 without a ChatGPT subscription: generation, editing and eight formats, on a balance shared with the whole catalogue.
- Hunyuan Image on Beemm VisionGenerate with Tencent's Hunyuan Image: text-to-image and editing, output up to 4K, eight formats — in a canvas that gathers the whole catalogue.
- AI image generator: every image modelModels of the same kind on one canvas, with the cost shown before each run.
- The full catalogueEvery image and video model available in the canvas.