Which LoRAs work here?

LoRAs trained for FLUX.1 [dev]; Hugging Face lists more than a thousand community LoRAs built for it. Pass a direct download link to the .safetensors file. SDXL, Z-Image or Qwen-Image LoRAs will not transfer.

Can I use uncensored LoRAs?

Yes. The weights are not reviewed and the output is not filtered, and in our tests LoRAs trained for unfiltered content rendered. At our strongest probe level the model held back part of the scene three times out of three, so for those requests Z-Image Turbo LoRA or Qwen Image 2512 LoRA is the better fit.

Can I use the images commercially?

Black Forest Labs' FLUX.1 [dev] license lets you use generated images for personal, scientific and commercial purposes. The license of each LoRA you load is set by its author, so check its model card.

What does guidance do?

It sets how strictly the image follows the prompt, from 0 to 20. The default 3.5 suits most LoRAs; realism LoRAs often look more natural a little lower, around 2 to 3, and very high values push toward a glossy, over-sharpened look.

Can I edit an existing picture?

Not on this family; it is text-to-image only. To restyle a picture with a LoRA, use the edit endpoint of Z-Image Turbo LoRA or Qwen Image 2512 LoRA, with a LoRA trained for that family's base.

How big are the images, and does the size change the price?

About one megapixel: 1024 x 1024 at 1:1, 1344 x 768 at 16:9 and 1536 x 640 at 21:9. Every ratio costs the same, and loading LoRAs adds no separate charge.

FLUX.1 Dev LoRA API: Uncensored NSFW Images, Any FLUX LoRA

black-forest-labs/flux-1-dev-lora/text-to-image

FLUX.1 Dev LoRA text-to-image from Black Forest Labs' FLUX.1 [dev], loading up to three of your own LoRA files per request. Output is about one megapixel in nine aspect ratios as JPEG, PNG or WebP, with a guidance control from 0 to 20.

From $0.018/image

Run model View API

Input

Text to image

Prompt224/5000

Aspect ratio

Style LoRAs1/3

Midjourney Mix2

Prompt adherence

Sampling steps

Output format

JPEGPNGWebP

Seed

Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run

Example result

Animate

Edit

Split into layers

{
  "model": "black-forest-labs/flux-1-dev-lora/text-to-image",
  "input": {
    "aspect_ratio": "2:3",
    "loras": [
      {
        "path": "https://huggingface.co/strangerzonehf/Flux-Midjourney-Mix2-LoRA/resolve/main/mjV6.safetensors",
        "scale": 1
      }
    ],
    "guidance_scale": 3.5,
    "steps": 28,
    "output_format": "jpeg",
    "seed": 42,
    "prompt": "MJ v6, editorial fashion photograph of a beautiful woman in a red satin gown with a plunging neckline and an open back, descending marble stairs and glancing over her bare shoulder, dramatic side light, glossy magazine look."
  }
}

Introduction

FLUX.1 Dev LoRA Text to image makes one image from a written prompt on Black Forest Labs' FLUX.1 [dev], a 12-billion-parameter rectified flow transformer, with up to three of your own LoRA files loaded. Pick any of nine aspect ratios at about one megapixel, set guidance, and get back JPEG, PNG or WebP at one flat price. The platform adds no content filter to the prompt, the weights or the image.

Choose a mode by starting material

What you have	Mode
a written description and a FLUX LoRA	FLUX.1 Dev LoRA Text to image
a picture to restyle	FLUX.1 Dev LoRA Image editing

Key features & highlights

The biggest LoRA library in open image models

Hugging Face lists more than a thousand LoRAs built on FLUX.1 [dev]: realism, fashion, film stocks, cartoons, characters and unfiltered styles. If you found a look you like, it is very likely a FLUX LoRA, and it runs here as a plain API call.

Guidance you can adjust

Guidance sets how strictly the image follows the prompt, from 0 to 20. The default 3.5 suits most LoRAs, and realism LoRAs often look more natural a little lower.

Uncensored with custom weights

The weights are never reviewed and the output is never filtered. In our tests unfiltered LoRAs rendered, though at the strongest level we tested the model held back part of the scene in all three tries.

Request parameters

Request parameters for FLUX.1 Dev LoRA · Text to Image
Parameter	Required	Description
`prompt`	Yes	Start with the LoRA's trigger phrase, then describe subject, setting and light in full sentences. FLUX follows long, specific prompts well. at least 1 characters · at most 5000 characters
`aspect_ratio`	No	Nine options from 21:9 to 9:21, all about one megapixel and all the same price. `1:1` · `16:9` · `9:16` · `4:3` · `3:4` · `3:2` · `2:3` · `21:9` · `9:21` · Default: `1:1`
`loras`	No	Up to three {path, scale} entries trained for FLUX.1 [dev]; scale 0 to 4, 1 is neutral. Strong style LoRAs may need a little more than 1. Default: `[]` at most 3 items
`guidance_scale`	No	How strictly to follow the prompt, 0 to 20. The default 3.5 suits most LoRAs; realism LoRAs often look better slightly lower. `≥ 0` · `≤ 20` · Default: `3.5`
`steps`	No	Denoising steps, 1 to 50 (default 28). More steps take longer to generate. `≥ 1` · `≤ 50` · Default: `28`
`output_format`	No	JPEG by default, PNG or WebP if you need them. Does not change the price. `jpeg` · `png` · `webp` · Default: `jpeg`
`seed`	No	Keeps a set of images comparable while you adjust guidance or one LoRA scale. `≥ 0` · `≤ 2147483647`

Pricing

$0.018 / image

Billed per image at one flat price for every aspect ratio. Loading LoRAs adds no separate charge.

Use cases

Run a LoRA you found

Paste the download link, set the scale, add the trigger phrase to your prompt. No training, no hosting and no GPU: the LoRA you saw on a model hub becomes a plain API call.

Editorial and fashion stills

Realism and magazine LoRAs give believable light, fabric and skin for gowns, portraits and night scenes, ready for mood boards and campaign drafts.

Stylised series

Cartoon and illustration LoRAs keep a whole set in one look across nine aspect ratios, from square posts to wide banners.

Explore key use-case tutorials

Prompt guide & pro tips

Trigger first, then full sentences

FLUX reads long prompts well. Start with the trigger phrase from the LoRA's model card, then describe subject, setting, clothing and light in complete sentences rather than a list of tags.

Lower guidance for realism

If a realism LoRA looks waxy or over-sharpened, drop guidance from 3.5 towards 2.5 and compare with the same seed. Higher guidance follows the prompt more literally but costs natural texture.

Raise weak styles one step at a time

When a style barely shows, raise its scale by 0.2. Our cartoon example needed 1.2. At high scales, check faces and hands closely; they are the first things to distort.

Keep text out of signs

FLUX tends to write words on any sign or screen in the frame, and they are often misspelled. Leave signs out of the prompt unless the exact wording matters.

API quick start

# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
  -H "Authorization: Bearer $SPICY_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
  -d '{
  "model": "black-forest-labs/flux-1-dev-lora/text-to-image",
  "input": {
    "aspect_ratio": "1:1",
    "output_format": "jpeg",
    "prompt": "A ceramic teapot on a linen cloth, soft window light, muted palette"
  }
}'

# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
  -H "Authorization: Bearer $SPICY_API_KEY"

See the API reference for authentication, the complete schema, polling, and error handling.

Related models and next tasks

Missing the model you need?

Tell us what to add and where you would use it. Requests go to the people who decide what gets onboarded next.

Run the FLUX LoRA you already found, as one API call

FLUX.1 Dev LoRA runs Black Forest Labs' 12-billion-parameter FLUX.1 [dev] with up to three of your own LoRA files per request. More than a thousand community LoRAs on Hugging Face are built for this base, from film realism and magazine finishes to cartoon styles and characters, so the look you want often exists as a file already. Paste its link, set the scale, write the prompt: nine aspect ratios at about one megapixel, a guidance control, one flat price per image and no platform filter. One thing to know up front: at our strongest content probe level, FLUX held part of the scene back.

Open the playground

Run the FLUX LoRA you already found, as one API call

LoRAs per request: Up to 3, each {path, scale}; scale 0 to 4, and 1 is neutral
Community library: More than a thousand LoRAs on Hugging Face built for FLUX.1 [dev]
Guidance: 0 to 20, default 3.5; realism LoRAs often look better around 2 to 3
Frame and price: About one megapixel in nine ratios, one flat price per image; LoRAs add nothing
Endpoints: Text to image only; for edits, Z-Image Turbo LoRA or Qwen Image 2512 LoRA
Content filter: None; at our strongest probe level the model held part of the scene back

Library: The look you want may already be a fileWith more than a thousand FLUX LoRAs on Hugging Face, the finish you are after has often been trained already: film realism, glossy editorial, cartoon styles, characters. Here each one is a link and a scale inside a request, with nothing to upload first and no review queue.
Second dial: Guidance works alongside the scaleGuidance, from 0 to 20 with a default of 3.5, sets how strictly the image follows the prompt. Realism LoRAs often look more natural a little lower, around 2 to 3, while very high values push toward a glossy, over-sharpened look.
Limits: Where FLUX holds backAt our strongest content probe level the model held back part of the scene in all three tries, even with a LoRA trained for that level loaded. Such LoRAs do render and nothing is filtered, but for the strongest requests Z-Image Turbo LoRA and Qwen Image 2512 LoRA both rendered that probe as written.

Getting a FLUX LoRA to look right on FLUX.1 Dev LoRA

Realism LoRAs: spend the prompt on light

Our diner example pairs a Super Realism LoRA with a late-night booth by a rainy window: two men leaning into each other under a warm pendant lamp. The prompt spends its words on light and posture, the LoRA supplies the photographic texture, and the result reads like a photograph. FLUX follows long, specific prompts well, so write full sentences.

Strong styles may need a scale above 1

Our long-limb cartoon LoRA was too faint at 1, so it ran at 1.2 and the whole frame turned into a cartoon illustration: the red umbrella, the wet cobblestones, the long-limbed look of the style. Start at 1 and move in steps of 0.2; the scale goes up to 4.

Put the trigger phrase first
Most FLUX LoRAs name a trigger phrase on their model card, such as "Super Realism" or "MJ v6". Put it at the very start of the prompt, then describe the picture. The card is also the best guide to the strength and prompt style that file expects.
Only FLUX.1 [dev] weights
LoRAs trained for SDXL, Z-Image or Qwen-Image will not transfer, and the reverse holds for FLUX files on those families. If a file seems to do little here, check which base its card names before you change anything else.
Public links, and the licence travels with the file
Use a direct link that downloads the .safetensors file; gated repositories answer with HTTP 401 and the request fails before any image is made. Black Forest Labs' FLUX.1 [dev] license lets you use generated images for personal, scientific and commercial purposes, while each LoRA's license is set by its author, so read its model card.

Who uses FLUX.1 Dev LoRA

Fashion and editorial teams
A magazine finish from one trigger
The "MJ v6" trigger and a Midjourney-style LoRA give FLUX a glossy editorial finish, here a red satin gown on marble stairs in hard side light. Keep the LoRA and the seed per campaign and the whole set shares that finish.
Lifestyle and travel campaigns
Warm, natural light from a realism LoRA
XLabs' realism LoRA at 0.9 with the default guidance of 3.5 gave warm golden-hour light on the water and on a man walking out of the surf. For a more natural look, try guidance a little lower before you add more adjectives.
Teams that collect LoRAs
A library of looks behind one request shape
Realism, editorial, cartoon: each look in this sheet came from a different community LoRA through the same request, with only the path, the scale and the trigger changing. Keep a list of files you trust, each pinned to an exact link, and switching look becomes a one-line change.
Dating, romance and social apps
Couples of every kind, one request shape
A Super Realism LoRA and two women laughing on a hotel balcony at dusk, one hand resting on the other's shoulder. Same-sex couples, mixed couples and friends are one prompt line apart: the LoRA carries the photographic texture, the prompt decides who is in the frame. When a frame later needs a restyle or a new camera angle, send it to the edit endpoint of Z-Image Turbo LoRA or Qwen Image 2512 LoRA.

From a model card to an image with FLUX.1 Dev LoRA

Read the model card
Note the base the file was trained on, its trigger phrase and any suggested strength. If the base is not FLUX.1 [dev], the file will not help here.
Copy the file link, not the page
Use the direct download URL of the .safetensors file from a public repository. A link that needs a login fails the request.
Write the prompt in the LoRA's own words
Trigger phrase first, then subject, setting and light in full sentences.
Start at scale 1 and guidance 3.5
Fix the seed, then move the scale in steps of 0.2 if the style is too faint or too strong, and lower guidance toward 2 or 3 for a more natural realism look. Change one dial at a time.
Settle the ratio, then run the set
Nine ratios from 21:9 to 9:21 at about one megapixel, all at one price, so draft in the shape you will publish.

FLUX.1 Dev LoRA prompts worth copying

A realism LoRA with its trigger

For a realism LoRA whose card names "Super Realism" as the trigger. The prompt spends its words on light and posture and leaves texture to the file.

Super Realism, two friends in their thirties laughing in a red vinyl diner booth at 2am, rain on the window behind them, a warm pendant lamp overhead, two coffee cups on the table, shallow depth of field

An editorial finish

For a Midjourney-style LoRA triggered by "MJ v6". Start at the default guidance of 3.5 and adjust from there.

MJ v6, editorial fashion photograph of a woman in a tailored emerald suit on a marble staircase, glancing back over her shoulder, dramatic side light, glossy magazine look

A cartoon style that needs more scale

For a strong cartoon LoRA. If the style is faint at 1, raise the scale to 1.2 as our example did.

Long toons, cartoon illustration with long stretched limbs and bold outlines, a man in a yellow raincoat splashing through puddles on a cobblestone street at night, street lamps and wet reflections

More FLUX.1 Dev LoRA questions

Start from the file, not the prompt

Find the FLUX.1 [dev] LoRA whose look you want, read its card for the trigger and the strength, and pass the direct file link. Write the trigger first and the scene in full sentences, start at scale 1 and guidance 3.5, and move one dial at a time with the seed fixed. When a frame needs editing, or a request goes further than FLUX will render, hand it to Z-Image Turbo LoRA or Qwen Image 2512 LoRA.