--- title: CivitAI API — Civitai Generator, Models & LoRAs by API description: CivitAI API: 30,000+ Civitai checkpoints and LoRAs by model_id, or import any model by URL. About $0.0047/image on the $47 plan, card billing, no Buzz. url: https://modelslab.com/civitai canonical: https://modelslab.com/civitai type: website component: Business/CivitAI generated_at: 2026-10-13T06:49:01.126317Z --- CivitAI API — Run Any Civitai Model or LoRA --- 30,000+ Civitai checkpoints and LoRAs are already imported: call any of them by model\_id, add a LoRA with lora\_model, or import a new model by URL. Standard is $47/month for 10,000 calls, about $0.0047 per image, billed by card with no Buzz. Search [Generate with CivitAI Models](https://modelslab.com/register) [Upload Model](https://modelslab.com/dashboard/upload-model/details) 30,000+ Civitai models imported 99.9% API Uptime REST API Access Last updated Sep 23, 2026 · By ModelsLab Engineering What the CivitAI API does --- The CivitAI API on ModelsLab runs Civitai community models from your code. 30,000+ Civitai models are already imported, 28,100 LoRAs and 2,290 checkpoints, and each one is called by its `model\_id`. Anything that is not imported yet can be added with one POST of its download URL. It replaces the Civitai on-site generator for developers: the same checkpoints and LoRAs, a REST endpoint with webhooks, and a card-billed monthly plan instead of Buzz. Standard is $47/month for 10,000 API calls, about $0.0047 per image. Generate with any Civitai model via API --- Imported Civitai checkpoints are called by model\_id on /api/v6/images/text2img. ### Python Python ``` 1import requests 2 3response = requests.post( 4 "https://modelslab.com/api/v6/images/text2img", 5 json={ 6 "key": "YOUR_MODELSLAB_API_KEY", 7 "model_id": "epicrealism-xl-v3-leo", 8 "prompt": "portrait of a lighthouse keeper, overcast light, 85mm", 9 "negative_prompt": "blurry, low quality", 10 "width": 1024, 11 "height": 1024, 12 "samples": 1, 13 "webhook": "https://your-app.example.com/hooks/image", 14 }, 15) 16 17print(response.json()) # output URLs, or an id for /api/v6/images/fetch/{id} ``` ### curl bash ``` 1curl https://modelslab.com/api/v6/images/text2img \ 2 -H "Content-Type: application/json" \ 3 -d '{ 4 "key": "YOUR_MODELSLAB_API_KEY", 5 "model_id": "anything-xl-xl", 6 "prompt": "a knight resting under a cherry tree, anime style", 7 "width": 1024, 8 "height": 1024 9 }' ``` Use a Civitai LoRA via API --- Base checkpoint in model\_id, LoRA in lora\_model, weight in lora\_strength, trigger word in the prompt. ### One LoRA (Python) Python ``` 1import requests 2 3response = requests.post( 4 "https://modelslab.com/api/v6/images/text2img", 5 json={ 6 "key": "YOUR_MODELSLAB_API_KEY", 7 "model_id": "sd-xl-v1-0-vae-fix", 8 "lora_model": "tshirtdesignredmond-t-shirt-design-lora-v2-0", 9 "lora_strength": 0.7, 10 "prompt": "TshirtDesignAF, a retro mountain sunset badge, flat colours", 11 "width": 1024, 12 "height": 1024, 13 }, 14) 15 16print(response.json()) ``` ### Two LoRAs (curl) bash ``` 1curl https://modelslab.com/api/v6/images/text2img \ 2 -H "Content-Type: application/json" \ 3 -d '{ 4 "key": "YOUR_MODELSLAB_API_KEY", 5 "model_id": "sd-xl-v1-0-vae-fix", 6 "lora_model": "stickers-redmond-stickers-lora-v1-0-sdxl,midjourney-mimic-v2-0", 7 "lora_strength": 0.6, 8 "prompt": "Stickers, a cheerful fox mascot, white border rounded-lg", 9 "width": 1024, 10 "height": 1024 11 }' ``` Import a Civitai model by URL --- For a checkpoint or LoRA that is not imported yet. Do it once; then call it by your model\_id. ### Import a LoRA (Python) Python ``` 1import requests 2 3response = requests.post( 4 "https://modelslab.com/api/v3/load_model", 5 json={ 6 "key": "YOUR_MODELSLAB_API_KEY", 7 "url": "https://civitai.com/api/download/models/YOUR_VERSION_ID", 8 "model_id": "my-brand-style-lora", 9 "model_name": "Brand style LoRA", 10 "model_image": "https://example.com/preview.png", 11 "model_category": "stable_diffusion_xl", 12 "model_subcategory": "lora", 13 "model_format": "safetensors", 14 "revision": "fp16", 15 "force_load": "no", 16 "model_visibility": "private", 17 "hf_upload": "no", 18 }, 19) 20 21print(response.json()) ``` ### The parameters, step by step On generation, `model\_id` is the base checkpoint. `lora\_model` is the LoRA id, or several ids separated by commas when you stack LoRAs. `lora\_strength` sets the weight; 0.6 to 0.8 is a common start. Put the LoRA’s trigger word in the prompt, exactly as the Civitai model page describes it. On import, `url` is the Civitai download link of the model version and `model\_id` is the name you will call it by; it must be unique. `model\_category` matches the base model: `stable\_diffusion`, `stable\_diffusion\_xl`, `flux` or `z\_image`. Add `model\_subcategory: lora` for a LoRA. `model\_format` is `safetensors`, `ckpt` or `diffusers`, and `model\_visibility` is `private` or `public`. ### What to expect on the first call A newly imported model has to download and load on a GPU before it serves requests. If a LoRA is not loaded yet next to your base checkpoint, the API starts loading it and answers with a message to retry in about five minutes. After that, calls go straight through. Models that people call every day are already loaded. Civitai generator (Buzz) vs a ModelsLab plan --- | | Civitai on-site generator | ModelsLab CivitAI API | |---|---|---| | How you pay | Buzz in several colours; part of generation paid in crypto since April 2026 | Card, monthly plan from $21 | | Which models you can run | Depends on the Buzz type and membership tier | Any imported model, plus any model you import | | Access from code | Built around the website | REST API, webhooks, Python and JavaScript examples | | Cost at 10,000 images | Varies by Buzz type and model | $47 on Standard (about $0.0047 each) | Civitai details from its own announcements: civitai.com/articles/21109 (Oct 2025), civitai.com/articles/28369 (16 Apr 2026) and civitai.com/articles/33269 (mid-2026). ModelsLab plans as listed on /pricing on 2026-09-23. ### CivitAI API pricing per image Every call needs an active plan, billed by card. Basic is $21/month for 3,250 API calls and Standard is $47/month for 10,000, which works out at about $0.0047 per image. The $149/month Open Source Unlimited plan does not meter self-hosted checkpoints and LoRAs, which includes every imported Civitai model, and allows 15 requests in parallel. Related pages --- [### CivitAI alternative Comparing platforms rather than looking for an API? The full comparison lives here.](https://modelslab.com/civitai-alternative) [### Flux LoRA training Train your own LoRA and call it with lora\_model, exactly like an imported Civitai LoRA.](https://modelslab.com/flux-lora-training) Get Expert Support in Seconds We're Here to Help. --- Want to know more? You can email us anytime at Chat with support[View Docs](https://docs.modelslab.com) Civitai has a public REST API for browsing and downloading models, and its own generator is paid in Buzz. The ModelsLab CivitAI API is a generation API: 30,000+ Civitai checkpoints and LoRAs are already loaded on GPUs, you call them by `model_id` on /api/v6/images/text2img, and you pay by card on a monthly plan. POST https://modelslab.com/api/v6/images/text2img with your ModelsLab `key`, the model's `model_id` (for example `epicrealism-xl-v3-leo`), a `prompt`, `width` and `height`. The response has the image URL, or an id to poll at /api/v6/images/fetch/{id}. Every imported model page shows its `model_id`. Send the base checkpoint in `model_id`, the LoRA id in `lora_model` and its weight in `lora_strength` (0.6 to 0.8 is a common start), and put the LoRA's trigger word in the prompt. `lora_model` accepts several ids separated by commas when you stack LoRAs. POST https://modelslab.com/api/v3/load_model with `url` (the Civitai download link), your own `model_id`, `model_name`, `model_image`, `model_category` (`stable_diffusion`, `stable_diffusion_xl`, `flux` or `z_image`), `model_format` (`safetensors`, `ckpt` or `diffusers`), `revision`, `force_load`, `model_visibility` and `hf_upload`. Add `model_subcategory: lora` for a LoRA. Plans are billed by card: Basic is $21/month for 3,250 API calls and Standard is $47/month for 10,000, about $0.0047 per image. The $149/month Open Source Unlimited plan does not meter self-hosted checkpoints and LoRAs and allows 15 requests in parallel. There is no Buzz and no crypto payment. Not always. A new checkpoint has to download and load on a GPU first, and if a LoRA is not yet loaded next to your base model the API answers with a message to retry in about five minutes while it loads. Models that people call every day are already loaded. For developers, yes: the same community checkpoints and LoRAs, called from code, billed by card. If you are comparing platforms rather than looking for an API, the CivitAI alternative page covers the differences in more detail. Yes. Flux LoRA training trains a LoRA from your own images and makes it available on the same generation endpoint, so you call it with `lora_model` like any imported Civitai LoRA. [ AI Image](https://modelslab.com/models?feature=imagen) [ AI Video](https://modelslab.com/models?feature=videogen) [ AI 3D](https://modelslab.com/models?feature=threedverse) [ AI Audio](https://modelslab.com/models?feature=audiogen) [ LLM APIs](https://modelslab.com/models?feature=llmaster) [ Training AI](https://modelslab.com/models?feature=training) ## Frequently Asked Questions ### Does Civitai have an API? Civitai has a public REST API for browsing and downloading models, and its own generator is paid in Buzz. The ModelsLab CivitAI API is a generation API: 30,000+ Civitai checkpoints and LoRAs are already loaded on GPUs, you call them by `model_id` on /api/v6/images/text2img, and you pay by card on a monthly plan. ### How do I generate an image with a Civitai model via the API? POST https://modelslab.com/api/v6/images/text2img with your ModelsLab `key`, the model's `model_id` (for example `epicrealism-xl-v3-leo`), a `prompt`, `width` and `height`. The response has the image URL, or an id to poll at /api/v6/images/fetch/{id}. Every imported model page shows its `model_id`. ### How do I use a Civitai LoRA via the API? Send the base checkpoint in `model_id`, the LoRA id in `lora_model` and its weight in `lora_strength` (0.6 to 0.8 is a common start), and put the LoRA's trigger word in the prompt. `lora_model` accepts several ids separated by commas when you stack LoRAs. ### How do I import a Civitai model that is not on ModelsLab yet? POST https://modelslab.com/api/v3/load_model with `url` (the Civitai download link), your own `model_id`, `model_name`, `model_image`, `model_category` (`stable_diffusion`, `stable_diffusion_xl`, `flux` or `z_image`), `model_format` (`safetensors`, `ckpt` or `diffusers`), `revision`, `force_load`, `model_visibility` and `hf_upload`. Add `model_subcategory: lora` for a LoRA. ### How much does the CivitAI API cost? Plans are billed by card: Basic is $21/month for 3,250 API calls and Standard is $47/month for 10,000, about $0.0047 per image. The $149/month Open Source Unlimited plan does not meter self-hosted checkpoints and LoRAs and allows 15 requests in parallel. There is no Buzz and no crypto payment. ### Is a newly imported model ready at once? Not always. A new checkpoint has to download and load on a GPU first, and if a LoRA is not yet loaded next to your base model the API answers with a message to retry in about five minutes while it loads. Models that people call every day are already loaded. ### Is ModelsLab a replacement for the Civitai generator? For developers, yes: the same community checkpoints and LoRAs, called from code, billed by card. If you are comparing platforms rather than looking for an API, the CivitAI alternative page covers the differences in more detail. ### Can I train my own LoRA instead? Yes. Flux LoRA training trains a LoRA from your own images and makes it available on the same generation endpoint, so you call it with `lora_model` like any imported Civitai LoRA. --- *This markdown version is optimized for AI agents and LLMs.* **Links:** - [Website](https://modelslab.com) - [API Documentation](https://docs.modelslab.com) - [Blog](https://modelslab.com/blog) --- *Generated by ModelsLab - 2026-10-13*