portweights

You fine-tuned the model. We'll serve it.

Serving a model yourself means renting a GPU, paying for it while it sits idle, and setting up inference. Portweights handles all of that. Upload your LoRA adapter and get a private, OpenAI-compatible endpoint and API key.

curl $BASE_URL/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3-8b-1a2b3c4d",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

How it works

  1. 01

    Sign in with Google

    One click. No forms.

  2. 02

    Upload your adapter

    Pick the two adapter files, or point us at a Hugging Face repo. Private repos work too.

  3. 03

    Call your endpoint

    We give you a base URL and an API key. Use them with curl or any OpenAI SDK.

What you can deploy

A LoRA adapter trained on Qwen/Qwen3-8B, rank 64 or lower. You need two files:

Base models

  • Qwen3-8B Available
  • Qwen3-4B-Instruct-2507Coming soon

Regions

  • India Available
  • USComing soon