What is the difference from FLUX.2 Klein Base 4B LoRA?

Both run Black Forest Labs' 4B Klein model. This one is the distilled build, made to finish in a few steps: in our tests it was about four times faster and followed edit instructions more reliably. Base 4B is the undistilled build its makers recommend for training; it is slower, gives more varied results and shows style LoRAs more strongly.

Which LoRAs work here?

Files trained for FLUX.2 [klein] 4B or Klein Base 4B. LoRAs from both builds loaded in our tests. LoRAs for FLUX.1 or the 9B Klein models load silently and have no effect.

Can I use uncensored LoRAs for NSFW images?

Yes. Weights are not reviewed and results are not filtered. In our tests an unfiltered LoRA from a model hub brought out Strong-level results that the model softened on its own; the Very strong level stayed toned down even then.

Why did my LoRA request fail?

Usually the link needs a login (gated Hugging Face files answer with HTTP 401) or points to a repository page instead of the file. Use a link that downloads the .safetensors file itself. Failed requests are not charged.

My LoRA loaded but nothing changed. Why?

It was probably trained for another base model, or it needs its trigger phrase. Check the LoRA page for the trigger and put it first. Style LoRAs can also be faint on this distilled build; try a scale of 1.5 to 2.

Can I stack several LoRAs?

Yes, up to three per request, each with its own scale. A fourth entry is rejected. Loading LoRAs adds no separate charge.

Does a bigger size cost more?

No. Every aspect ratio and both size tiers cost the same per image, in text to image and in edits.

Am I charged for a failed request?

No. Failed requests are not charged.

FLUX.2 Klein 4B LoRA API: Uncensored NSFW Images, Your LoRAs

black-forest-labs/flux-2-klein-4b-lora/text-to-image

FLUX.2 Klein 4B LoRA text to image on Black Forest Labs' FLUX.2 [klein] 4B, loading up to three of your own LoRA files per request. Nine aspect ratios at about one megapixel or 2048 pixels on the long side, one flat price per image.

From $0.016/image

Run model View API

Input

Text to image

Prompt562/5000

Aspect ratio

Resolution

Style LoRAs1/3

Mikkoani Klein4b

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-2-klein-4b-lora/text-to-image",
  "input": {
    "aspect_ratio": "16:9",
    "resolution": "2k",
    "loras": [
      {
        "path": "https://huggingface.co/mikkoph/mikkoani-klein4b/resolve/main/klein4b-mikkoani.safetensors",
        "scale": 1
      }
    ],
    "seed": 63327,
    "prompt": "@mikkoani illustration. Blue hour in the gravel lot of a roadside diner in the desert. An adult Latina woman in her thirties in high-waisted denim shorts and a red flannel shirt knotted at the waist, bare shoulders, sits on the concrete curb beside an adult South Asian man in his thirties in a white t-shirt and an open denim jacket. They share one strawberry milkshake with two straws, leaning in close, her hand on his knee. The diner windows glow warm behind them, a tall blank pole sign glowing plain turquoise against the cobalt sky, no lettering anywhere."
  }
}

Introduction

FLUX.2 Klein 4B LoRA Text to image makes one image from a written prompt on Black Forest Labs' FLUX.2 [klein] 4B, the fast distilled 4-billion-parameter model released under Apache 2.0, with up to three of your own LoRA files loaded. Pick any of nine aspect ratios at about one megapixel or 2048 pixels on the long side, at one flat price per image. Nothing is filtered.

Choose a mode by starting material

What you have	Mode
a written description and a LoRA	FLUX.2 Klein 4B LoRA Text to image
one to three pictures to change or combine	FLUX.2 Klein 4B LoRA Image edit

Key features & highlights

Results in seconds

The distilled build finished most of our text-to-image requests in about 4 seconds of model time, which makes it practical to try several seeds and LoRA scales for one idea.

LoRAs from both Klein 4B builds

LoRAs trained on FLUX.2 [klein] 4B and on Klein Base 4B both loaded in our tests. Files for FLUX.1 or for the 9B Klein models load without an error and do nothing.

Uncensored with custom weights

Weights are never reviewed and results are never filtered. On its own the model tones down Strong-level prompts; with an unfiltered LoRA loaded, those came through as written. The Very strong level stayed toned down.

Request parameters

Request parameters for FLUX.2 Klein 4B LoRA · Text to Image
Parameter	Required	Description
`prompt`	Yes	Up to 5,000 characters. Start with the LoRA trigger phrase, then describe subject, clothing, place and light in full sentences. at least 1 characters · at most 5000 characters
`aspect_ratio`	No	Nine shapes from 21:9 to 9:21, 1:1 by default. Every ratio costs the same. `1:1` · `16:9` · `9:16` · `4:3` · `3:4` · `3:2` · `2:3` · `21:9` · `9:21` · Default: `1:1`
`resolution`	No	1k (default) is about one megapixel; 2k is 2048 pixels on the long side. Both tiers cost the same. `1k` · `2k` · Default: `1k`
`loras`	No	Up to three {path, scale} entries trained for FLUX.2 [klein] 4B or Base 4B. Default: `[]` at most 3 items
`seed`	No	A number from 0 to 2147483647. Reuse it to compare one LoRA or one scale at a time; it keeps results close but does not promise identical pixels. `≥ 0` · `≤ 2147483647`

Pricing

$0.016 / image

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

Use cases

Running a LoRA you found

Paste the download link from Hugging Face or Civitai, set the scale and put the trigger first. No training, no hosting and no GPU: a LoRA you saw on a model hub becomes a plain API call that returns in seconds.

Illustrated series

Comic, colored-pencil and painted LoRAs keep a set of images in one style, from square posts to 21:9 banners. Fix the seed per layout and change only the subject line between requests.

Fast drafts for a bigger model

Try compositions, outfits and LoRA scales quickly here at the 1k tier, then render the chosen frame at 2k or hand the prompt to a slower model for the final image.

Explore key use-case tutorials

Prompt guide & pro tips

Trigger first, then full sentences

Start with the trigger phrase from the LoRA page, then describe subject, clothing, place and light. Name the lens and the grade for photographic results.

Push faint styles to 1.5 or 2

Style LoRAs show less on this distilled build. Raise the scale step by step and compare with the same seed; at 4 the image fell apart in our test.

Compare against no LoRA

Run the same prompt and seed without the LoRA. If the two results look the same, the file was made for another base model, such as FLUX.1 or the 9B Klein models: those load without an error and do nothing here.

Keep signs out of the frame

The model writes garbled words on signs and labels. Describe signs as blank or leave them out unless the text matters.

Use direct, public links

The link must download the .safetensors file itself. A Hugging Face file link or a Civitai download link works; a repository page or a gated file that needs a login fails the request, which is then not charged.

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-2-klein-4b-lora/text-to-image",
  "input": {
    "aspect_ratio": "1:1",
    "prompt": "A ceramic teapot on a linen cloth, soft window light, muted palette",
    "resolution": "1k"
  }
}'

# 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.

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.

LORA IMAGE API

FLUX.2 Klein 4B LoRA API at a glance

FLUX.2 [klein] 4B with your own LoRAs: the fast, distilled 4B model for text to image and editing, and nothing filtered.

Parameters: 4B; Black Forest Labs' distilled FLUX.2 [klein] 4B, Apache 2.0 open weights.
LoRAs per call: 3; Up to three weight files per image, each with its own scale from 0 to 4.
Typical render: ~4 s; Median text-to-image request time across our tests; edits took about 7 seconds.
Source images: 3; Edit or combine up to three pictures in one request.
Largest side: 2048 px; Nine aspect ratios at about one megapixel or 2048 pixels on the long side.
Content filter: None; No filter on prompts, weights or results.

Results

What you can make with FLUX.2 Klein 4B LoRA

Style
A comic look from one LoRA
An illustration LoRA with its trigger at the start: a couple sharing a milkshake on the curb of a desert diner at blue hour, drawn with ink lines and flat colour at 16:9 and 2k.
Text to image with an illustration LoRA at 1, 16:9 at 2k.
Restyle
A photo redrawn in colored pencil
An edit with a colored-pencil LoRA loaded: a daytime diner photo came back as a blue-hour drawing, with her face, curls and yellow dress kept.
Image edit with a colored-pencil LoRA at 1, aspect ratio auto.
Base model
Film photographs without any LoRA
No LoRA loaded: the model alone draws a film-grain portrait at a diner counter at blue hour, at 2k from a few sentences.
Text to image without a LoRA, 3:4 at 2k.

More clips made with FLUX.2 Klein 4B LoRA

Image edit with two sources: a portrait and a dress, 3:4 at 2k.
Text to image with a colored-pencil LoRA at 1, 2:3.
Image edit with one source, aspect ratio auto.
Image edit with a painted-fairytale LoRA at 1, reframed to 16:9.
Text to image without a LoRA, 21:9 at 2k.
Text to image with a painted-fairytale LoRA at 1, 9:16.

Content test

Is FLUX.2 Klein 4B LoRA uncensored? What we tested

We ran each level on our live setup and checked every clip frame by frame. A pass means the content was actually on screen, not just that the request went through.

Content levels tested on FLUX.2 Klein 4B LoRA, by mode
Level	Image editing	Text to image
MildThe lightest prompts in the test	RendersTested Oct 9, 2026	RendersTested Oct 9, 2026
ModerateWhere filtered models start to refuse	RendersTested Oct 9, 2026	RendersTested Oct 9, 2026
StrongThe middle of the scale	RendersTested Oct 9, 2026	RendersTested Oct 9, 2026
Very strongWhere most models start to hold back	Not tested yet	Not tested yet
MaximumThe top of the scale	RendersTested Oct 9, 2026	RendersTested Oct 9, 2026

Renders: on screen as written. Softened: generated, but milder than what was written. Refused: rejected. Not tested yet: we have not run that level on that mode.

How each mode behaves

All modes: Use LoRAs trained for FLUX.2 [klein] 4B or Klein Base 4B. A LoRA made for FLUX.1 or for the 9B Klein models loads without an error and changes nothing, and one community file in our tests turned the whole frame into grey noise without an error. A link that needs a login or points to a web page fails the request. Up to three LoRAs, scale 0 to 4; at 4 the image fell apart.
All modes: Nothing is reviewed or refused, but the model itself is conservative: without an unfiltered LoRA it often tones down Strong-level prompts, and the Very strong level came back toned down even with one loaded. Both Maximum-level prompts rendered.
All modes: There is no negative prompt, step count or guidance control. Signs and labels in the frame often come out as misspelled text, so leave them out unless you need them. Output is one JPEG per request, up to 2048 pixels on the long side.
All modes: Every person you describe must be an adult. You are responsible for the lawfulness of what you generate and for having the rights to any source image and LoRA you use.

Use cases

What to build with the FLUX.2 Klein 4B LoRA API

Running community LoRAs fast
Paste a Hugging Face or Civitai link to a Klein 4B LoRA and get results in seconds, without hosting a GPU.
Illustration sets
Colored pencil, comic and painterly LoRAs keep a whole series in one look across nine aspect ratios.
Restyling photos
Edit a photo with a style LoRA loaded and it comes back redrawn in that style with the person and setting kept.
Unfiltered LoRAs on an unfiltered model
Weights are never reviewed and results are never filtered. The base model holds back on its own, so load the LoRA your work needs.

Prompts

FLUX.2 Klein 4B LoRA prompt examples you can copy

Paste one into the playground as it is, then change one thing at a time.

Text to image
Diner counter at blue hour
Blue hour through the long windows of an empty roadside diner at the edge of the desert. An adult East Asian woman in her late twenties sits sideways on a chrome counter stool in a mint-green waitress dress with the top two buttons undone and the sleeves pushed up, one knee drawn up, a slice of cherry pie and a glass coffee pot on the counter beside her. She looks over her shoulder at the camera with a slow, knowing smile. Cobalt dusk outside, warm pink light from the pie case inside, chrome reflections. 35mm film photograph, Kodak Portra 400 colour, shallow depth of field, soft grain.
Why it works. No LoRA. Film stock, lens and the two light sources are named, so the model returns a grainy photograph rather than a clean render.
Try this prompt
Image editing
Satin dress on the motel walkway
Dress the woman from image 1 in the dusty-rose satin slip dress from image 2, barefoot, one thin strap slipping off her shoulder, her hair let down. Place her leaning on the iron railing of a second-floor motel walkway at blue hour, looking back at the camera with a quiet, alluring smile; a row of numbered doors and the desert under a cobalt sky behind her. Keep her face. Realistic photo, warm porch light.
Why it works. Two sources: a studio portrait as image 1 and a dress on a hanger as image 2. The dress transferred exactly; the face shifted a little, so compare it with your source.
This mode starts from your own image: add it in the playground before you run.
Try this prompt
Text to image
Coloured-pencil ice machine
Colored pencil hyperdetailed realism, an adult Black man in his early thirties in the open-air breezeway of a single-storey desert motel at blue hour, filling a plastic bucket at a humming ice machine. A cream short-sleeve bowling shirt hangs open over a white ribbed undershirt, loose khaki trousers, a room key on a green plastic tag hooked on one finger. He glances back over his shoulder with a relaxed, smouldering half smile. A row of turquoise doors recedes behind him, the sky fading from cobalt to orange over a black mountain ridge.
Why it works. A colored-pencil LoRA with its trigger phrase first. An earlier seed gave him a frown and a signature in the corner; a new seed and a clear expression fixed both.
Try this prompt

Pricing

FLUX.2 Klein 4B LoRA API pricing

One image starts at $0.016.

The playground shows the exact price for your settings before you run. Failed or timed-out tasks are usually not charged. Some models charge for content refusals as stated on their model page; check those billing terms before submitting.

See every price tier

Get started

How to use the FLUX.2 Klein 4B LoRA API in three steps

Step 1
Pick LoRAs made for Klein 4B
Look for LoRAs whose base model is FLUX.2 [klein] 4B or Klein Base 4B; both loaded in our tests. LoRAs for FLUX.1 or the 9B Klein models load without an error and change nothing.
Step 2
Trigger first, then sentences
Start with the trigger phrase from the LoRA page, then describe subject, clothing, place and light in full sentences. There is no negative prompt, so say what you want rather than what to avoid.
Step 3
Push a weak style a little
Style LoRAs show less on this fast distilled build than on Base 4B. If a look barely appears at 1, try 1.5 to 2 and compare with the same seed. At 4 the image fell apart.

Open the playground Get an API key

Parameter reference and code samples are in the API and Readme tabs at the top of this page.

Why SpicyAPI

Why run FLUX.2 Klein 4B LoRA on SpicyAPI

No filter from us
We add no content filter of our own. What the model renders is what you get back.
Tested, not guessed
Content labels come from real generations we ran and checked, with the test date on the page.
Pay per result, in dollars
No subscription and no credits. Failed or timed-out tasks are usually not charged. Some models charge for content refusals as stated on their model page; check those billing terms before submitting.
One key for every model
One API key and balance cover image, video, audio and chat models, so you can explore more models from the same account.

FAQ

FLUX.2 Klein 4B LoRA API FAQ

About FLUX.2 Klein 4B LoRA

FLUX.2 Klein 4B LoRA puts Black Forest Labs' fast, Apache 2.0 FLUX.2 [klein] 4B behind your own LoRAs: text to image in nine aspect ratios up to 2048 pixels and editing from up to three source images, with up to three weight files per request. LoRAs load without review, and nothing is filtered.