How much does Anything v4.0 cost?

It runs on a T4 and bills by GPU time. A typical 6-second generation is about $0.0012, roughly 833 runs per dollar, which makes it the cheapest image model in the Hormouz catalogue. Current rates are shown in the Hormouz dashboard.

What is the largest image it can produce?

1024x768 or 768x1024. Memory limits mean one side must stay at or below 768, so a full 1024x1024 is not available on this model. The default is 512x512.

Can it edit an existing image?

No. Anything v4.0 is text-to-image only. For image-to-image or inpainting in an anime style, use Proteus v0.3, which accepts an input image and a mask.

How should I write prompts for it?

Use danbooru-style tags rather than prose — short comma-separated descriptors such as "1girl, long hair, school uniform, outdoors, bokeh". A negative prompt helps remove recurring artefacts.

anything-v4.0: image generation model

Anything v4.0 is a highly detailed anime-style Stable Diffusion model, and one of the cheapest image models available — roughly $0.0012 per run. Served asynchronously through Hormouz from inside Iran.

Pricing:: ~$0.000225 / sec (Nvidia T4 GPU)
Type:: image

Image generation

Anything v4.0 is a well-established anime-style Stable Diffusion model, built for high-quality, highly detailed anime and illustration output. It responds to the danbooru-style tag prompts the anime community uses, so short comma-separated tags work better here than long prose descriptions.

It is text-to-image only — there is no image input and no inpainting. You control size, the number of images, the sampler, denoising steps and guidance scale, and a seed makes a result reproducible.

Output dimensions are chosen from a fixed set between 128 and 1024 pixels, defaulting to 512x512. Memory limits cap the largest output at 1024x768 or 768x1024, so one side must stay at or below 768. You can request one or four images per run.

This is by some distance the cheapest model of its kind here. It runs on a T4 and bills by GPU time, and a typical 6-second generation works out around $0.0012 — roughly 833 runs per dollar. Hormouz provides stable access from inside Iran, with no foreign credit card or circumvention tooling required.

Quickstart

# Create the prediction, then poll until succeeded
curl -s https://api.hormouz.net/v1/predictions \
  -H "Authorization: Bearer ng-YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"version":"cjwbw/anything-v4.0","input":{"prompt":"A beautiful sunset"}}'

Specs

Vendor	Cjwbw
Type	Image
Context window	—
Pricing	~$0.000225 / sec (Nvidia T4 GPU)
API	predictions

Code example

cURL

# Create the prediction, then poll until succeeded
curl -s https://api.hormouz.net/v1/predictions \
  -H "Authorization: Bearer ng-YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"version":"cjwbw/anything-v4.0","input":{"prompt":"A beautiful sunset"}}'

Python

import requests
BASE = "https://api.hormouz.net/v1"
H = {"Authorization": "Bearer ng-YOUR_API_KEY", "Content-Type": "application/json"}
r = requests.post(f"{BASE}/predictions", headers=H,
    json={"version": "cjwbw/anything-v4.0", "input": {"prompt":"A beautiful sunset"}})
print(r.json()["id"])

JavaScript

const BASE = "https://api.hormouz.net/v1";
const res = await fetch(`${BASE}/predictions`, {
  method: "POST",
  headers: { Authorization: "Bearer ng-YOUR_API_KEY", "Content-Type": "application/json" },
  body: JSON.stringify({ version: "cjwbw/anything-v4.0", input: {"prompt":"A beautiful sunset"} }),
});
const { id } = await res.json();

Buy the anything-v4.0 API, paid in rial

Buy credit with an Iranian bank card; no foreign card or account needed.
One Hormouz key reaches anything-v4.0 and 600+ other models, and you pay only for what you use.
No VPN and no risk of a banned account; just change the base URL to api.hormouz.net/v1.

Create an account

Use via Hormouz

Connect to this model and 600+ others with one API key, then see usage down to each request.

Request company access

FAQ

How much does Anything v4.0 cost?: It runs on a T4 and bills by GPU time. A typical 6-second generation is about $0.0012, roughly 833 runs per dollar, which makes it the cheapest image model in the Hormouz catalogue. Current rates are shown in the Hormouz dashboard.
What is the largest image it can produce?: 1024x768 or 768x1024. Memory limits mean one side must stay at or below 768, so a full 1024x1024 is not available on this model. The default is 512x512.
Can it edit an existing image?: No. Anything v4.0 is text-to-image only. For image-to-image or inpainting in an anime style, use Proteus v0.3, which accepts an input image and a mask.
How should I write prompts for it?: Use danbooru-style tags rather than prose — short comma-separated descriptors such as "1girl, long hair, school uniform, outdoors, bokeh". A negative prompt helps remove recurring artefacts.

Updated 2026-10-13