People who look photographed
Skin keeps its pores, wrinkles and uneven tone instead of the waxy finish older FLUX versions were known for. Portraits, street style and creator-style shots read as photos.
FLUX.3 · by Black Forest Labs
FLUX AI image generator
FLUX is the image model family from Black Forest Labs. The newest, FLUX.3 Image, is known for natural skin and faces, text that comes out as written, and edits that change one thing and leave the rest alone. Describe an image below, or add a photo to edit, and it opens in Image Studio on FLUX.3.
Example image
Opens in Image Studio with the model you picked. You see the price in credits before anything runs.
What FLUX is known for
Text comes out as written
Put the exact words in quotes and say where they go. Covers, labels, menus and posters come back spelled right, in several languages; reviewers call typography the biggest jump over earlier FLUX versions.
Edits that touch one thing
FLUX.3 Image Edit changes what you name (a jacket, a sign, the background) and leaves the rest of the picture alone, and it takes up to ten reference images.
A fast, cheap tier for drafts
FLUX.3 Klein is the small, quick version of the same model. Use it to find the picture, then render the keeper on FLUX.3 Image.
Where it's weaker
- No seed and no negative prompt: the same prompt gives a different image each run.
- Saying what you don't want backfires ("no crowds" can add crowds). Describe what you do want: "an empty street".
- Hands holding small objects and tangled poses still go wrong sometimes.
- On Affogato it renders at 1K. The model can go to 4K elsewhere, but takes minutes there.
- On public leaderboards it is the strongest FLUX, not the strongest image model overall: GPT Image and Nano Banana rank higher.
- Klein's small or distant text is often unreadable; use FLUX.3 Image for anything with words.
- Maker
- Black Forest Labs, Germany · FLUX.3 Image released October 2026
- Output on Affogato
- 1K, one image per run
- Aspect ratios
- 9:16 · 3:4 · 1:1 · 4:3 · 16:9 and more, from 1:2 to 21:9
- Text in images
- Yes, multilingual; quote the exact words
- Edit references
- Up to 10 images (FLUX.3 Image Edit)
- Leaderboard
- Top FLUX model on the Artificial Analysis text-to-image arena (Nov 2026)
Made with FLUX on Affogato
Every image below was generated with the prompt shown, unedited. Load any prompt into the generator and make your own version.
Fashion on a personTrench coat on a Seoul crossing
Prompt
Vertical street-style fashion photo: a woman in her late twenties with a cropped platinum bob walks toward the camera across a zebra crossing in Seoul, wearing an oversized camel trench coat over a black turtleneck, wide charcoal trousers and white leather sneakers, a small burgundy shoulder bag. Mid-stride, one hand in her pocket, looking just past the lens. Overcast midday light, shallow depth of field, blurred taxis and pedestrians behind her. Shot on a 50mm lens, natural skin texture, editorial magazine look.
Text in the imageA magazine cover, every line spelled right
Prompt
A vertical magazine cover. Masthead at the top in tall white serif capitals: "NORTHBOUND". Below it, a young Black man with short twists in a rust-colored corduroy jacket stands on a windy coastal cliff, looking into the camera with a calm half-smile. Cover lines on the left in clean sans-serif: "The Slow Travel Issue", "48 hours in the Faroe Islands", "Why we stopped flying". Bottom right a small issue label: "No. 12 · Winter 2026". Natural overcast light, sharp print-quality photograph.
UGC lookA creator's chili-crisp post
Prompt
Vertical UGC-style phone photo: a smiling woman in her thirties with curly dark hair, in a mustard apron, holds a glass jar of homemade chili crisp up to the camera in her small apartment kitchen. The hand-written jar label reads "Mira's Chili Crisp — HOT". Steam rises from a pan behind her, warm window light, a little cluttered countertop with garlic, scallions and a wooden spoon. Looks like a real creator's Instagram post, natural skin, slight phone-camera grain.
Skin and ageA fisherman at dawn, in a navy cap
Prompt
Close-up vertical portrait of a fisherman in his seventies on a wooden pier at dawn: deep sun-worn wrinkles, white stubble, salt-dried skin, pale blue eyes squinting slightly, a faded navy knit cap and an oilskin jacket with water droplets on the collar. Soft low golden light from the left, the sea out of focus behind him. 85mm lens, f/2, every pore and hair sharp, true-to-life color, no retouching.
Frozen motionOut of the blocks
Prompt
Vertical sports photo: a female sprinter explodes out of the starting blocks on a red running track, low angle from the side, body angled forward, one knee driving up, arms pumping, tension in every muscle, beads of sweat on her temple, determined eyes. Stadium lights and a blurred empty grandstand behind. Frozen motion at 1/4000 s, crisp detail on her plain black spikes and the track's white lane line. Plain unbranded black kit, no logos anywhere.
Several people at onceFive friends, one toast
Prompt
Wide candid photo of five friends in their twenties laughing around a small table on a rooftop at golden hour, city skyline behind them. One raises a glass mid-toast, one covers her mouth laughing, one leans back in his chair, one takes a phone photo of the others, one hands a plate of tacos across. Different ethnicities, natural mixed clothing, string lights just switching on. Every face, hand and glass rendered clearly. 35mm documentary style, warm natural light.
Signs and menusThe barista and the chalkboard
Prompt
Square photo inside a small neighborhood café: a tattooed barista with a grey beanie leans on the counter and points at a chalkboard menu behind him, laughing. The chalkboard reads, in neat hand-drawn chalk lettering: "TODAY" on top, then "Oat Flat White 4.5", "Cardamom Bun 3", "Iced Hojicha 5". Morning light, espresso machine steam, real café clutter, candid documentary style.
HandsSixty years, two rings
Prompt
Square close-up: an older couple's hands, both in their eighties, holding each other on a white tablecloth at their grandchild's wedding. Thin wrinkled skin, visible veins, her pearl ring and his worn gold band, a sprig of eucalyptus beside them, a blurred champagne glass in the background. Soft window light, film photography look, true anatomy, five fingers on each hand.
Which FLUX to pick
Start here
FLUX.3 Image
Most images
The flagship: people, products with labels, posters and covers. Start here.
10 credits per image
Open with FLUX.3 ImageOption 2
FLUX.3 Klein
Drafts and quick ideas
The small, fast FLUX.3. Try compositions here, then re-run the one you like on FLUX.3 Image.
2 credits per image
Open with FLUX.3 KleinOption 3
FLUX.2 [max]
Older FLUX workflows
The previous FLUX flagship, with web-grounded detail. Keep it for prompts tuned on FLUX.2; start new work on FLUX.3.
14 credits per image
Open with FLUX.2 [max]Add a photo in the generator above and FLUX.3 edits it instead (FLUX.3 Image Edit). FLUX.2 [pro], FLUX 1.1 [pro] and the FLUX.1 Kontext edit models are still in Image Studio's model menu. Credits are read from the live catalog; the Standard plan ($24/month) includes 2,000 credits, about 200 FLUX.3 Image images.
Prompting FLUX well
- Lead with the subject, then the place, the style, the camera and the light. The order changes the framing.
- Short prompts work: FLUX.3 fills in detail. Add words only for what you want to control.
- Put any text in double quotes and say where it goes and how it looks ("masthead in tall white serif capitals").
- Describe what you want, never what you don't. "An empty promenade" works; "no people" can add people.
- Name a lens and light for photo realism ("85mm, f/2, soft window light"); ask for natural skin texture by name.
- For edits, name the thing to change and say what stays: "Make her jacket red, keep everything else."
- Draft on FLUX.3 Klein, then render the keeper on FLUX.3 Image.
From prompt to image
Describe the picture
Write it in the box at the top of this page, or start from an example. Any prompt in the gallery loads with one click.
Sign in and land in Image Studio
Your prompt, format and the FLUX version you picked carry over. Nothing runs until you press Generate.
Generate, then pick a keeper
The credit price shows under the Generate button. Run a few takes; every run is a new image.
Edit or animate it
Send the image to FLUX.3 Image Edit to change one detail, or to Video Studio to bring it to life.
Questions about FLUX
What is FLUX best at?
Photoreal people with natural skin, text that comes out spelled as written (covers, labels, menus, posters) and edits that change one detail while keeping the rest. It is weaker on hands holding small objects, and it has no negative prompt.
Which FLUX version should I use?
FLUX.3 Image for finished pictures, FLUX.3 Klein for fast drafts and FLUX.3 Image Edit to change a photo you already have. The older FLUX.2 and FLUX.1 Kontext models are still there if a workflow depends on them.
Is FLUX.3 better than Nano Banana or GPT Image?
Not overall: on the Artificial Analysis arena GPT Image and Nano Banana rank above it. FLUX.3 is the strongest FLUX and a good pick for photographic people and exact text, and it costs fewer credits per image than most flagship image models in our catalog.
Can FLUX put readable text in an image?
Yes, FLUX.3 Image is one of its strengths. Put the exact words in double quotes and describe where they go. Keep long text on FLUX.3 Image rather than Klein.
Is FLUX free on Affogato?
No. Every image costs credits, shown before you generate. Credits come with a monthly plan and unused credits roll over up to the plan's cap.
Can I use FLUX images commercially?
Yes. Black Forest Labs grants commercial use of images made through its API, and paid Affogato plans include a commercial license and no watermark, within our terms of service.
Does FLUX make videos too?
Yes. FLUX.3 also has video models, which you will find in Video Studio's model menu. This page covers the image models.
Can I call FLUX from code?
Yes. The FLUX models on Affogato are also available through the Affogato API and MCP server, at the same credit price.
Keep exploring
Facts checked November 15, 2026. Sources: Black Forest Labs: FLUX 3 announcement · Black Forest Labs: FLUX 3 Image Klein · BFL docs: release notes · BFL docs: FLUX 3 Image overview · BFL docs: prompting basics · BFL help: commercial use of API output · The Decoder: FLUX 3 Image launch · Artificial Analysis: text-to-image arena · MindStudio: FLUX 3 Klein tests · fal: FLUX 3 Image API