SDXL-Lightning, Stable Diffusion XL’den damıtılmış, yalnızca birkaç adımda yüksek kaliteli 1024px görseller üreten yıldırım hızında bir text-to-image modelidir.
SDXL Lightning 4-step is a text-to-image AI model by bytedance. On ModelRunner it runs through a REST API or via MCP from any AI assistant, at about $0.00215033 per image.
POST https://queue.modelrunner.run/bytedance/sdxl-lightning-4step
cURL
# Submit a request to the queue. Input fields go at the top level of the
# body. The optional reserved "metadata" object holds your own flat string
# tags — stored on the request, never sent to the model; filter later with
# GET https://queue.modelrunner.run/requests?metadata=<url-encoded JSON>.
curl -X POST https://queue.modelrunner.run/bytedance/sdxl-lightning-4step \
-H "Authorization: Key $MRUN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"seed": 104,
"width": 1280,
"height": 1024,
"prompt": "rainy evening street with an Art Deco café storefront; the neon sign clearly reads \u0027LUMINA\u0027, wet pavement reflections…",
"scheduler": "K_EULER",
"num_outputs": 1,
"guidance_scale": 0,
"negative_prompt": "worst quality, low quality, cluttered scene, illegible text, misspelled text, blurry, watermark, logo",
"num_inference_steps": 4,
"disable_safety_checker": false,
"metadata": {
"project": "my-project"
}
}'
# → { "request_id": "...", "status_url": "...", "response_url": "..." }
# Poll status_url until "COMPLETED", then fetch the result
curl "https://queue.modelrunner.run/bytedance/sdxl-lightning-4step/requests/$REQUEST_ID/status" \
-H "Authorization: Key $MRUN_API_KEY"
curl "https://queue.modelrunner.run/bytedance/sdxl-lightning-4step/requests/$REQUEST_ID" \
-H "Authorization: Key $MRUN_API_KEY"
JavaScript
import { modelrunner } from "@modelrunner/client";
const result = await modelrunner.subscribe("bytedance/sdxl-lightning-4step", {
input: {
"seed": 104,
"width": 1280,
"height": 1024,
"prompt": "rainy evening street with an Art Deco café storefront; the neon sign clearly reads 'LUMINA', wet pavement reflections…",
"scheduler": "K_EULER",
"num_outputs": 1,
"guidance_scale": 0,
"negative_prompt": "worst quality, low quality, cluttered scene, illegible text, misspelled text, blurry, watermark, logo",
"num_inference_steps": 4,
"disable_safety_checker": false
},
});
console.log(result);
Python
import os
import requests
headers = {"Authorization": f"Key {os.environ['MRUN_API_KEY']}"}
submitted = requests.post(
"https://queue.modelrunner.run/bytedance/sdxl-lightning-4step",
headers=headers,
json={
"seed": 104,
"width": 1280,
"height": 1024,
"prompt": "rainy evening street with an Art Deco café storefront; the neon sign clearly reads 'LUMINA', wet pavement reflections…",
"scheduler": "K_EULER",
"num_outputs": 1,
"guidance_scale": 0,
"negative_prompt": "worst quality, low quality, cluttered scene, illegible text, misspelled text, blurry, watermark, logo",
"num_inference_steps": 4,
"disable_safety_checker": false
},
).json()
# Poll submitted["status_url"] until "COMPLETED", then:
result = requests.get(submitted["response_url"], headers=headers).json()
Input parameters
Input parameters of SDXL Lightning 4-step
Name
Type
Required
Description
prompt
string
no
Input prompt Default: "self-portrait of a woman, lightning in the background".
Use SDXL Lightning 4-step from Claude & Cursor (MCP)
Point Claude Code, Claude Desktop, Cursor, or any MCP client at the ModelRunner MCP server and SDXL Lightning 4-step becomes a tool your assistant can call directly — it authorizes via OAuth (no API key in config) and runs this model with the run_model tool using the endpoint bytedance/sdxl-lightning-4step.
claude mcp add --transport http modelrunner https://mcp.modelrunner.run/mcp
Then ask your assistant, for example: “Run bytedance/sdxl-lightning-4step on ModelRunner to generate image”. MCP setup guide.
Model Detayları
Model Detayları
# SDXL-Lightning
**SDXL-Lightning**, hem **hız** hem de **kalite** için tasarlanmış, son teknoloji bir text-to-image üretim modelidir; şaşırtıcı derecede az adımda **yüksek çözünürlüklü 1024px görseller** üretebilir.
[`stabilityai/stable-diffusion-xl-base-1.0`](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0) temeli üzerine kurulmuştur ve hızlı ama güçlü görsel sentezi için *Progressive Adversarial Diffusion Distillation* yöntemini kullanır.
## Özellikler
- 🚀 1, 2, 4 ve 8 adımlık damıtılmış modellerle **yıldırım hızında inference**.
- **2, 4 ve 8 adımlık** modeller olağanüstü görsel kalitesi sunar.
- **1 adımlık** model deneyseldir; uç verimliliğin sınırlarını araştırır.
- 🧩 **Model formatları**:
- **Tam UNet checkpoint’leri** – en iyi görsel kalitesini verir.
- **LoRA checkpoint’leri** – hafiftir ve başka temel modellere kolayca uygulanır.
## Open Source
SDXL-Lightning’i, topluluğu en ileri hızlı diffusion teknolojisiyle güçlendirme araştırmamızın bir parçası olarak tamamen open source yayımlıyoruz. Ayrıntılar için makalemize bakın:
**SDXL-Lightning: Progressive Adversarial Diffusion Distillation**.
