Instant BYOM

Instant BYOM turns a public Hugging Face model into a tenant-scoped Parel model ID. There is no dedicated GPU to keep warm and no idle charge.

Looking for image generation? See Image Models for Flux, SD 3.5, Qwen-Image, and SDXL through /v1/images/generations.

Flow

1. Validate the Hugging Face model.
2. Import it into your account.
3. Use the returned model ID in chat completions.

import httpx
from openai import OpenAI

headers = {"Authorization": "Bearer pk-dev-YOUR_KEY"}

httpx.post(
    "https://api.parel.cloud/v1/my-models/validate-hf",
    headers=headers,
    json={"huggingface_id": "Qwen/Qwen2.5-7B-Instruct"},
).raise_for_status()

created = httpx.post(
    "https://api.parel.cloud/v1/my-models/import-hf",
    headers=headers,
    json={
        "huggingface_id": "Qwen/Qwen2.5-7B-Instruct",
        "display_name": "My Qwen 7B",
    },
).json()

client = OpenAI(api_key="pk-dev-YOUR_KEY", base_url="https://api.parel.cloud/v1")
response = client.chat.completions.create(
    model=created["id"],
    messages=[{"role": "user", "content": "Merhaba"}],
)

Endpoints

Endpoint	Description
POST /v1/my-models/validate-hf	Check access, architecture, and serving path
POST /v1/my-models/import-hf	Register the model to your account
GET /v1/my-models	List imported models
GET /v1/my-models/{id}	Read model details
DELETE /v1/my-models/{id}	Remove an imported model