Uncensored AI image generator API: five models and their price per image

An uncensored AI image generator API runs prompts that mainstream image APIs filter out, and returns the picture from one HTTP request. Five uncensored image models run on one UplinkSwitch key, next to the uncensored LLM API models and everything else in the catalog. Each is priced per image, and every price on this page is read live from the catalog.

Developers shopping for an NSFW image API want three numbers: the price per image, the size of the file, and how long the call takes. The table has all three for each model, measured on 2026-10-19.

You can try any of the five without code, from a prompt box in the console. The API call is one POST with a model id and a prompt. Both are below.

Most uncensored AI image generator sites are web apps with a prompt box. The console's Media page is that prompt box here, and it runs the same models the API does.

1. Sign in to the console and add credit, from $10. There is no subscription.
2. Create a key on the Keys page. Media runs as one of your keys, so each image is billed like an API call.
3. Open Media. Under Try a generation, type the prompt, pick one of the five under Image, pick the key, and press Run. The price per image shows next to the picker before you run it.
4. The Generations list below the panel shows the status, the cost, and a link to the file when it is done.

The five models answer POST /v1/images/generations the way OpenAI's image API does (OpenAI image generation guide). Send the model id and a prompt of up to 1,000 characters. The call waits for the image and answers with its URL.

curl https://api.uplinkswitch.com/v1/images/generations \
  -H "Authorization: Bearer $UPLINK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "chroma-uncensored",
    "prompt": "A lighthouse on a sea cliff at dusk, oil painting, warm light"
  }'

{
  "id": "gen_7d2e…",
  "created": 1792388160,
  "data": [
    { "url": "https://api.uplinkswitch.com/v1/files/2026-10-19/gen_7d2e…/0.png" }
  ]
}

The image is at data[0].url, on our own domain. Open it or download it like any file; it stays there for 30 days. The response headers carry the generation id in x-uplink-generation-id and the charge in x-uplink-charge-usd. Send "response_format": "b64_json" to get the image bytes in the JSON instead of a URL.

The OpenAI SDK works as it is, pointed at our base URL:

from openai import OpenAI

client = OpenAI(base_url="https://api.uplinkswitch.com/v1", api_key="sk-uplink-live-...")
result = client.images.generate(
    model="vision-image-uncensored",
    prompt="A lighthouse on a sea cliff at dusk, oil painting, warm light",
)
print(result.data[0].url)

Each model renders at its own size, listed in the table above. A size field and n: 1 are accepted, so existing OpenAI image code runs unchanged. One request makes one image: for four images, send four requests. The media API guide covers the second endpoint, POST /v1/generations, which answers at once with a generation id to poll.

Lustify SDXL Uncensored is an SDXL model, run for you. It costs $0.009 an image. In our test it returned a 1024x1024 PNG in about 9 seconds.

Pick it if your prompts are already written for SDXL models. Here it is one request, with no GPU to rent and no weights to download. Full details are on the Lustify SDXL API page.

Anime Image Uncensored makes anime and illustrated images at $0.018 an image. It returned a 1024x1024 PNG in about 5 seconds, the quickest answer of the five in our run.

Pick it for character art in anime and illustration styles, and for working one prompt over many quick drafts. Settings are on the Anime Image Uncensored model page.

Chroma Uncensored costs $0.003 an image and returned a 1536x1536 PNG in about 6 seconds. That is the largest output of the five, a size it shares with Vision Image Uncensored.

Pick it for large square images in volume: 1,000 of them come to $3.00. See Chroma Uncensored settings and price.

Vision Image Uncensored costs $0.003 an image. In our test it returned a 1536x1536 PNG in about 7 seconds.

Pick it when you want a second model at 1536x1536: send the same prompt to Vision and to Chroma, and keep the one you like. More on the Vision Image Uncensored page.

Vision Pro Image Uncensored is the Pro build of the Vision model, at $0.045 an image. It returned a 1024x1024 JPEG in 26 to 34 seconds in our test.

Pick it for finished images where a half-minute render fits the work, such as a final pass on a prompt you drafted on Vision or Chroma. Its settings are on the Vision Pro Image Uncensored API page.

The price is set when the request is accepted and charged then. A generation that fails is refunded in full, automatically. Credit is prepaid, so there is no invoice and no plan to cancel. Every other image model on the key is in the text-to-image price list, and video, speech and 3D are on the media models page.

The same key also makes uncensored video and edits: the uncensored AI video generator API (Wan 3.0 Video Uncensored) from $0.15 a second of output, and the uncensored image edit API (Seedream 5.0 Lite Edit Uncensored) at $0.0313 an edit.

Errors come back in OpenAI's shape, with the code in error.code. These are the ones an image request can meet:

The images endpoint waits up to 100 seconds for the file. Every model measured well inside that on 2026-10-19. If a call does run past it, the 504 carries the generation id in x-uplink-generation-id: read GET /v1/generations/{id} until the status is succeeded, and the file URL is in outputs.

Call the API from your server, never from code that runs in a browser: anyone holding the key spends your credit. In the console, a key can carry a list of the models it may call and a monthly spend limit. Give your image app a key limited to the five ids above, and a request for any other model comes back 403 scope_denied. If the key leaks, revoke it in the console and create a new one.

On 2026-10-19 we called each of the five models through POST /v1/images/generations with our own key and recorded the file type, its pixel size and how long the call took. Those are the Output and Time columns above.

Prices are not typed into this post. They are read from GET /v1/catalog/media each time the page is rebuilt, at most five minutes apart, so the $0.003 shown for Chroma Uncensored is today’s price. The chat models on the same key get a daily real-model check, with every verdict public.

Get a key: prepaid from $10, no subscription, 30-day price lock. Open the console.

What AI can draw uncensored?

On this key, five image models: Lustify SDXL Uncensored, Anime Image Uncensored, Chroma Uncensored, Vision Image Uncensored and Vision Pro Image Uncensored. Each runs prompts that mainstream image APIs filter out, within the two acceptable-use rules below, at $0.003 to $0.045 an image.

Is there an uncensored AI image generator with an API?

Yes. All five answer POST https://api.uplinkswitch.com/v1/images/generations with a model id and a prompt, in OpenAI's image format, and return a URL to the finished image. The same key runs every other model in the catalog.

What are the rules for uncensored images?

Two. No image may depict a minor: a prompt that puts a minor in the picture is refused with a 400 content_not_allowed error before anything runs, so nothing is charged. And no image of a real person’s likeness without that person’s consent.

Is uncensored AI free?

No. Each image bills from prepaid credit: Lustify SDXL Uncensored $0.009, Anime Image Uncensored $0.018, Chroma Uncensored $0.003, Vision Image Uncensored $0.003, Vision Pro Image Uncensored $0.045, live from the catalog. Credit starts at $10 with no subscription, and a generation that fails is refunded in full.

Which uncensored image model is fastest?

In our 2026-10-19 test, Anime Image Uncensored answered in about 5 seconds, Chroma Uncensored in about 6, Vision Image Uncensored in about 7, Lustify SDXL Uncensored in about 9 and Vision Pro Image Uncensored in 26 to 34.

Can I use the OpenAI SDK?

Yes. Set base_url to https://api.uplinkswitch.com/v1, use an UplinkSwitch key, and call images.generate with one of the five model ids. The image URL comes back in data[0].url.

Can one request make several images?

No. Each request makes one image, and n above 1 is refused with a 400 that says so. Send one request per image; each is billed as one image.

How long are the images kept?

Thirty days, at a URL on api.uplinkswitch.com. Download what you want to keep, or send response_format: "b64_json" to get the image bytes in the response instead.

Can the same key run uncensored chat models?

Yes. The key that calls these five image models also calls the uncensored and abliterated chat models on POST /v1/chat/completions, billed per token. They are listed on the Uncensored LLM API page.

How do I create uncensored AI images?

Send POST https://api.uplinkswitch.com/v1/images/generations with your key, a model id and a prompt, in OpenAI's image format; the response carries a URL to the finished image. On 2026-10-19, Lustify SDXL Uncensored answered in about 9 seconds at $0.009 an image; Anime Image Uncensored answered in about 5 seconds at $0.018 an image; Chroma Uncensored answered in about 6 seconds at $0.003 an image; Vision Image Uncensored answered in about 7 seconds at $0.003 an image.

What is the best uncensored AI image generator?

By price, Chroma Uncensored and Vision Image Uncensored at $0.003 an image. By output size, the 1536x1536 PNG from Chroma Uncensored and Vision Image Uncensored. By speed in our 2026-10-19 test, Anime Image Uncensored at about 5 seconds. Vision Pro Image Uncensored is the slowest and the dearest at $0.045 and 26 to 34 seconds. We have not scored image quality.

Can I make uncensored AI images without Stable Diffusion?

Yes. None of the 5 models here needs a local install: each runs behind the API and bills per image from prepaid credit. Lustify SDXL Uncensored is SDXL-based; you still install nothing.