OpenAI Playground

Before you begin, make sure you have an OpenAI account and have configured your API key.

export OPENAI_API_KEY="your_api_key_here"

Once in your account, click Playground in the top navigation to access the interface.

Playground Interface Overview

The Playground is organized into three main sections:

• Getting Started panel (left): Usage guidelines, best practices, and limitations.
• Presets toolbar (top): Quickly load common tasks like Classification, Chat, and more.
• Editor & Controls (center and right): Write prompts and adjust parameters such as temperature and max_tokens.

Classification Preset: Company Category Labeling

Select the Classification preset to label items by category. By default, you’ll see examples like Apple, Facebook, and FedEx. Let’s classify “Apple”:

It correctly labels Apple as a technology company.

Below the editor, click View Code to examine the underlying API request in languages such as Python, Node.js, or raw JSON. For example, here’s the Python snippet:

import os
import openai

openai.api_key = os.getenv("OPENAI_API_KEY")

response = openai.Completion.create(
    model="text-davinci-003",
    prompt="The following is a list of companies and the categories they belong to:",
    temperature=0,
    max_tokens=6,
    top_p=1,
    frequency_penalty=0,
    presence_penalty=0,
    stop=["\n"]
)
print(response.choices[0].text.strip())

Experimenting with Models & Parameters

Beyond presets, the Playground lets you switch between models—such as text-davinci-003 and gpt-3.5-turbo—and fine-tune parameters. Hover over any label to view its description and experiment to see how changes affect output.

Modes Overview

The Playground offers four modes to suit different tasks:

1. Complete Mode

In Complete mode, you get free-form text completions. For instance, ask for a tagline:

Write a tagline for an online bicycle shop.

Click Submit, and you might see:

The Right Bike for Any Ride!

2. Chat Mode

Switch to Chat to leverage conversational models like GPT-3.5 Turbo. For example, role-play a history professor:

• System prompt: “You are an expert in world history.”
• User prompt: “When did America become independent?”

Submit to receive a detailed response:

3. Insert Mode

Use Insert mode to tell the model where to fill in text. Example:

[Insert the body of an announcement email recognizing sales teams for their performance.]

Submit to generate a complete email section:

4. Edit Mode

In Edit mode, paste existing text and ask the model to improve or correct it:

Fix the grammar mistakes in the following paragraph:
"Last night i went to the movie blah blah blah..."

After submission, you’ll receive a polished version:

You now have an overview of the OpenAI Playground’s presets, models, modes, and parameter controls. Continue experimenting to uncover best practices and optimize your prompts.

Further Reading

• OpenAI API Documentation
• Playground Guide
• Model Reference