Welcome to Tune Studio docs.

Tune Studio is developer backend for your GenAI journey stitching the most used APIs together into a single platform. It is used to quickly ship apps, track performance and train models to build the best products. Integrates your keys for LLM APIs, Weights & Biases, HuggingFace, etc.

Tune Studio Enterprise can be privately hosted to work with your hybrid cloud strategy. For more questions reach out to sales.

Click here for API reference.

Signup

First thing you need to do is signup on Tune Studio, easiest way is to use Google. Using integrations and model APIs is free of cost for life. Finetuning and deployment requires your credit card on file.

  • Set up your account and make sure to confirm your email address
  • Copy the organisation ID from the settings page. You might need this for x-org-id header.
  • For creating a team, go to Members page and invite up to 3 members instantly. Want more seats, see how to add card on file.

Installing package

You can also install our python client from pypi like:

pip install tuneapi

It is still in beta and if you have any feature requests or come across a bug please report to our issue tracker.

Need help?

If you need any help you can reach out to us here:

  • Discord: good for community and code support
  • X: good for social engagements
  • LinkedIn: good for business partnerships

Quickstart

Tune Studio is fantastic for quickly building and deploying models. Here’s a quick guide to get you started. We will connect to OpenAI API and track logs to build a dataset.

Integrating API provider

For your first steps we will integrate OpenAI APIs and track logs to build a dataset. Here’s what you need to to do:

  1. Select OpenAI from the Integrations page and paste the API key
  2. To change the default behaviour go to model Settings -> Log API requests. For this demo it should be enabled.
  3. Click on Dashboard -> Open in Playground to open the connected model in the playground where you can use the UI to test ideas.
  4. Click on API button to open the API references. Tune Studio supports text completion, chat completion and function calling
  5. As you use the model your API logs table will start filling up which you can then open in playground to track usage.

Our chat is compatible with OpenAI, read more on their documentation. Here’s a quick breakdown of supported formats. Find keys here.

You can use the standard chat completion like this. Add x-tune-key, x-org-id and model.

curl -X POST \
  https://proxy.tune.app/chat/completions \
  -H "authorization: <Tune Studio API Key>" \
  -H "x-org-id: <Tune Studio Org ID>" \
  -H "Content-Type: application/json" \
  -d '{
    "temperature": 0.9,
    "messages": [
      {
        "role": "system",
        "content": "You are TuneStudio"
      },
      {
        "role": "user",
        "content": "Who are you"
      }
    ],
    "model": "<Tune Model ID>",
    "stream": false,
    "penalty": 0.2,
    "max_tokens": 100
  }'

For a more details read Integrations and Models.

Training your model

Training the model is super easy with Tune Studio, you can either use the UI or APIs. First thing you’ll need is to get the training data. Go to the model that you have been using and click on the Download in the API logs table to download a JSONL file of all the items.

See Finetuning and Deploy for more details.

More Examples

The best way to learn more about Tune Studio is to recreate examples:

Lots of APIs

Tune Studio has a ton of APIs to build your GenAI backend fast. Here’s a few to get you started:

  • ProxyAPI to use the models and integrations
  • ModelAPI for configuring and deploying different models
  • ThreadAPI to store conversations and chat logs in a structured way