Commissioned
Getting Started

Quick Start

Go from zero to a fine-tuned model in under five minutes.

This guide gets you from sign-up to chatting with your own fine-tuned model as fast as possible.

No machine learning experience is needed. If you can drag and drop a file, you can fine-tune a model.

Prerequisites

  • A web browser
  • Data you want to train on (documents, conversations, code — any text)
  • That's it

The 5-minute walkthrough

Create your account

Go to app.commissioned.tech and sign up with email, Google, or GitHub. No credit card required for the free tier.

Upload your data

From the dashboard, drag files into the upload area. Commissioned accepts:

  • JSONL / JSON — structured data, conversation logs
  • PDF — documents, papers, reports
  • TXT / Markdown — plain text, notes, articles

Files can be up to 5 GB each. Upload as many as you need.

Describe what you want

In the description field, write a plain-English explanation of your goal:

"I want a writing assistant that matches the tone and style of these blog posts"

"Create a support agent that can answer questions based on our help documentation"

This description guides how Commissioned cleans and structures your data for training.

Pick a base model

Select a base model from the dropdown:

ModelBest forTraining time
GPT-4.1 MiniGetting started, general use30–45 min
GPT-4.1Complex tasks, highest quality30–45 min
Gemini 2.5 FlashLong documents30–45 min
Qwen 3 8BFast training, self-hosting~5 min

If you're unsure, pick GPT-4.1 Mini — it's fast, capable, and free.

Click "Create a Custom Model"

Commissioned takes over from here:

  1. Parses and extracts text from your files
  2. Cleans and deduplicates the content
  3. Formats it for the target provider
  4. Submits the training job
  5. Monitors until completion

You'll get an email when it's ready. GPU models (Qwen) finish in ~5 minutes; cloud models take 30–45 minutes.

Start chatting

Once your model shows Succeeded, click Open in chat. You're now talking to a model trained on your data.

Try asking it something specific to your domain — you'll notice it understands your context without any prompting.

What's next

Now that you have a working model, you can:

On this page