Selecting Between OpenAI GPT models

OpenAI has released several models of the GPT (Generative Pre-trained Transformer) series, each with its own unique specifications and use cases. In this document, we will discuss the most optimal use cases for GPT-3.5 Turbo, GPT-3.5 16k, and GPT-4.

Choosing between OpenAI Base Models

• GPT-3.5-turbo: This model is the fastest and most cost effective model. It is suitable for general applications that do not require extensive context memory or complex function calling. A good base prompt paired with GPT-3.5 Turbo can produce decent results, such as the Wonderchat bot on our website. It has lower resource requirements than GPT-4, making it ideal for projects with limited resources. It can handle roughly 20 pages of text in a single request.
• GPT-3.5 (16k): This model is an updated version of GPT-3.5-turbo that can handle larger texts, making it more effective for applications that require processing substantial amounts of text data. It has a 16k context, which means it can understand and generate natural language or code with a larger context window.
  • Best suited for: Users with long custom prompts, PDF pages with content that is in a repetitive format
• GPT-4: This model has advanced function calling capabilities and a larger context window than GPT-3.5-turbo**2. It is the way to go if you need these capabilities. However, it comes at a higher price point compared to GPT-3.5-turbo, and its larger context window and enhanced capabilities come with increased computational resource needs2. It features an updated and improved model with function-calling capabilities3**.

In summary, if you are working on a budget or with limited resources, GPT-3.5-turbo is a suitable choice

• If your application doesn’t require extensive context memory or complex function calling, GPT-3.5-turbo will serve you well
• If you require your AI to have a large context window because you have a long custom instruction for your chatbot, GPT-3.5 Turbo (16k) is a good choice.
• If you need your chatbot to constantly cite accurate links or for your chatbot to do advanced maths or calculation, GPT-4 is the way to go