Create LoRA

POST /v2/loras

Multipart endpoint. The body must contain 15–30 image files, a name, and (for the default person subject kind) a gender and ethnicity. Returns 202 Accepted with the new LoRA's id and initial status — training continues in the background and you should poll GET /v2/loras/:id (or use a webhook) to track progress.

Body — `multipart/form-data`

Field	Type	Description
`images`	file[]	Required. Between 15 and 30 image parts (jpg / png / webp). Each ≤ 10 MB, total ≤ 100 MB. The field name `images` may be repeated.
`name`	string	Required. Display name for the LoRA (1–60 chars). Surfaced in your dashboard listings.
`subjectKind`	string	Optional. One of `person` (default), `style`, `object`.
`gender`	string	Required when `subjectKind=person`. One of `female`, `male`, `non-binary`. Baked into the captioning prompt and the trigger phrase to materially improve fidelity.
`ethnicity`	string	Required when `subjectKind=person`. Free-form (1–40 chars). The studio constrains it to `White`, `Black`, `East Asian`, `South Asian`, `Southeast Asian`, `Hispanic/Latino`, `Middle Eastern`, `Mixed`, but the API accepts any string.
`age`	string	Optional, free-form (1–40 chars). E.g. `"25"`, `"early 30s"`, `"around 40"`.
`steps`	integer	Optional. Number of training steps (500–3000, default 1000). Higher = better quality + slower + same flat 255cr cost.
`loraRank`	integer	Optional. LoRA rank (8–64, default 32).
`webhookUrl`	string	Optional HTTPS URL — receives a JSON POST when training reaches a terminal state.
`webhookEvents`	string	Optional comma-separated subset of `completed,failed`.

Response — `202`

{
  "id": "lora_8a4d2c…",
  "status": "PENDING",
  "imageCount": 18
}

Errors

Code	When
`400`	Wrong number of images, oversize file, unsupported mime, missing required fields.
`402`	Insufficient credits (the 2cr `lora-create` fee couldn't be deducted).
`429`	Team is at its concurrent training cap (`maxConcurrentTrainings`, default 10).

Example

curl -X POST https://api.apiframe.ai/v2/loras \
  -H "X-API-Key: afk_your_api_key_here" \
  -F "name=Alex – studio portraits" \
  -F "subjectKind=person" \
  -F "gender=male" \
  -F "ethnicity=White" \
  -F "[email protected]" \
  -F "[email protected]" \
  # … repeat for all 15-30 images
  -F "[email protected]"

import requests

files = [("images", open(f"photo{i:02d}.jpg", "rb")) for i in range(1, 19)]
response = requests.post(
    "https://api.apiframe.ai/v2/loras",
    headers={"X-API-Key": "afk_your_api_key_here"},
    data={
        "name": "Alex – studio portraits",
        "subjectKind": "person",
        "gender": "male",
        "ethnicity": "White",
    },
    files=files,
)
print(response.json())

import fs from "node:fs";

const form = new FormData();
form.append("name", "Alex – studio portraits");
form.append("subjectKind", "person");
form.append("gender", "male");
form.append("ethnicity", "White");
for (let i = 1; i <= 18; i++) {
  form.append("images", new Blob([fs.readFileSync(`photo${String(i).padStart(2, "0")}.jpg`)]), `photo${i}.jpg`);
}

const response = await fetch("https://api.apiframe.ai/v2/loras", {
  method: "POST",
  headers: { "X-API-Key": "afk_your_api_key_here" },
  body: form,
});
console.log(await response.json());

Body — multipart/form-data

Response — 202

Errors

Example

On this page

Body — `multipart/form-data`

Response — `202`