If you are new to OpenAI’s Custom GPTs, check out our article GPTs: One Step Closer to Wider AI Utilization.
Before May 2024, the main reason for choosing an Assistant over a Custom GPT was that free ChatGPT users only had access to the former. This has changed with the advent of GPT4o, which opened Custom GPT usage to every user, no matter if they are free or paid subscribers.
So, is Custom GPT now the king of OpenAI solutions, or is there still a reason to opt for an Assistant?
Let’s find out in this article.
Note: OpenAI provides a great comparison here, but this article offers a different perspective on the debate. Hope you enjoy both sources!
Both Assistants and Custom GPTs can be used for building chatbots that:
If you want to incorporate the chatbot into your application or website, you need to create an Assistant. Custom GPTs are only accessible through the ChatGPT interface.
If you want users’ conversations with the chatbot to remain private and not be part of OpenAI’s training dataset, you should choose to build an Assistant over a Custom GPT. Although Custom GPTs offer the possibility of excluding user threads from training purposes, this option is only available if you have uploaded a knowledge file to the GPT. This setting can be found at the bottom of the GPT’s configuration page as shown in the image below. If, however, the GPT relies only on instructions and not on any knowledge file, this option is not available.
The creation process for a Custom GPT and an Assistant is very similar. The main difference is that Custom GPTs can generate the required instructions through conversation with the GPT owner, whereas for Assistants the instructions should come directly from the creators.
On the other hand, Assistants provide more flexibility in solution settings. Creators can choose:
Although the creation process is very similar, the accessibility of the two solutions differ substantially.
As of May 2024, the usage of Custom GPTs doesn’t incur any cost to the GPT creators. Conversely, OpenAI measures the usage of Assistants through tokens and creators are charged accordingly. The price per token depends on the GPT model on which Assistants are built. Older OpenAI models are typically cheaper than the latest GPT models.
For more information about OpenAI prices, visit this help page.
Choosing between OpenAI Assistants and Custom GPTs ultimately depends on your specific needs and resources. Custom GPTs offer a user-friendly, no-code solution that is easily accessible through the ChatGPT interface, making them ideal for individuals and organizations looking for simplicity and quick deployment without development overhead. On the other hand, Assistants provide greater flexibility and control, allowing for deeper integration into your applications and websites, but they require some technical expertise to implement.
Consider your priorities regarding data privacy, creation process complexity, accessibility, and cost when making your decision. Both solutions have their unique advantages and can significantly enhance user interactions and automate various tasks.
Explore both options, experiment with their features, and choose the one that best fits your project's goals.
Happy building!