ChatGPT’s Model Picker: Back and More Complex

OpenAI recently brought back the model picker in ChatGPT, but navigating its options has become a bit more intricate. Let’s break down what’s new and how it impacts your experience.

Understanding the Return of the Model Picker

The model picker lets users select which underlying language model powers their ChatGPT interactions. It allows you to choose between different versions, potentially optimizing for speed, accuracy, or specific tasks. This feature disappeared for a while but has now returned, presenting some interesting choices.

Navigating the Options

Previously, the model selection was more straightforward. Now, users might find themselves facing options such as:

• GPT-3.5: The older, faster, and cheaper option. It’s suitable for general tasks and quick conversations.
• GPT-4: The more powerful, slower, and more expensive model, ideal for complex tasks requiring reasoning and creativity.
• Specific Purpose Models: Some users may see specialized models tuned for particular applications.

The Complexity Explained

The apparent complexity comes from several factors:

• Increased Model Variety: OpenAI offers more models, leading to a wider range of choices.
• Dynamic Availability: Model availability can change based on demand and other factors.
• A/B Testing: OpenAI likely runs A/B tests, exposing different users to various model configurations to optimize performance and gather feedback.