- Learn Prompting's Newsletter
- Posts
- Custom GPTs vs. Gemini Gems: Who Wins?
Custom GPTs vs. Gemini Gems: Who Wins?
I built with both OpenAI and Google's custom AI tools – here's my take on which one you should use.
Learn Prompting Newsletter
Your Weekly Guide to Generative AI Development
Comparing Custom GPTs and Gems
Learn the strengths and weaknesses of Custom GPTs and Gems and how they compare to each other.
Hey there,
Ever felt like you're drowning in a sea of AI tools, but none of them do exactly what you need? You're not alone. That's where custom chat experiences come in as a way to build your own AI assistant. Today, we're putting two of the biggest players head-to-head: Google's Gems and OpenAI's Custom GPTs.
A Quick Refresher: What Are Custom Chatbots?
As a quick refresher for everyone, custom GPTs and Gems are personalized versions of ChatGPT and Gemini that you create. They are customized through behavioral changes, additional knowledge bases (files & APIs), or completely redesigned experiences. Think of custom chatbots as a tailored version of ChatGPT or Gemini. Their goal is to create a unique AI experience or an AI expert with a single purpose.
Comparing GPTs and Gems
Now that we’ve had a refresher on what custom GPTs and Gems are, let's compare the two.
Creation Process & Ease of Use:
ChatGPT: Creating custom GPTs in ChatGPT is super simple. They have two main modes: “Create” and “Configure”. Create allows you to describe what you're looking for while chatting with ChatGPT. After you’ve shared the details, it will auto-generate a preview. You can continue to edit this preview using the Configure mode shown below. Configure mode differs from Create by breaking your GPT into more defined parameters. I’ll often start with Create to get my idea off the ground and then refine it using Configure.

This is the GPT configure page.
Gemini: Creating Gems has a very similar process as custom GPTs. There are a few key differences however, mainly there is no “Create” mode. This means that you’ll have to manually describe the Gem you want. As you can see from the screenshots, the Gem building interface is much more barebones than the custom GPTs. While Gemini is ultimately looking for all the same information, they don’t neatly break it up to help guide creation. They do however have an AI-powered feature that will re-write and expand your description to help create a better Gem.

This is the Gem creation page.
Capabilities & Tools:
ChatGPT: ChatGPT offers a number of built-in tools that you can enable. Under the “Capabilities” section seen below, you have total control over what your GPT is capable of. Beside those important controls, you can also manually create conversation starters. This is great because it allows you to directly help your users get started.

The “Configure” menu helps make the GPT creation process easier.
Gemini: Unfortunately Gems don’t offer the same level of customization. While you can expand the knowledge base with up to 10 unique sources, you can’t disable search or image generation. While having these tools by default isn’t really an issue in most cases, it is a shame that you can’t easily toggle them off.
Knowledge & Data Integration:
There are two main ways to expand the knowledge base of your custom GPT, you can either upload files or you can connect to an API.
ChatGPT: Each custom GPT has a knowledge base limit of 20 files. Apart from this you can connect to external APIs to give your custom GPT greater flexibility.
Gemini: Your Gems have a smaller limit of just 10 files that can be referenced. This is odd because the Gemini models have larger context windows so I’d have thought that this limit would be significantly higher.
This is one of the most important elements of custom chat experiences. Once created, being able to share your project with users and finding other people’s creations is essential.
ChatGPT: Both discovering new custom GPTs and sharing your creations is super simple within ChatGPT. The GPT marketplace allows you to see and search through ALL public GPTs. You can search for keywords or explore certain areas like “Writing” or “Education”.
The GPT marketplace contains all public GPTs.
Sharing your creation is also quick and easy. Just click “Share” in the top right corner and select how you want to share your GPT.

There are 3 unique choices for sharing GPTs: private, link based, and public.
Gemini: Unfortunately Gems are very limited in terms of sharing and discovery. You currently cannot share Gems with other users. Because of this, there is no Gem marketplace. This is perhaps the greatest weakness of Gems.
Access:
ChatGPT: Unfortunately you need to have a paid plan to create your own custom GPTs. The Plus Plan ($20/month) grants you this access. You DO NOT need a paid plan to use custom GPTs, just to create your own.
Gemini: This is one of the only areas where Gemini clearly beats out ChatGPT and is the biggest advantage of Gems. Anyone with a Gmail account can create Gems for free.
Are Custom Chatbots Still Relevant in 2025?
Custom chatbots like GPTs have been around for years now and with the sheer number of AI tools currently available, the question needs to be asked, are custom chatbots like GPTs and Gems still relevant?
Yes, I’d say that they are. Having the ability to create powerful custom experiences that are highly tailored to your needs is still necessary. While some custom chatbot use cases are being filled by more professional and standalone tools, there are still more that aren’t. You also need to remember that while there may be a standalone writing or education tool on the market, it won’t be as customizable as a custom chatbot.
The Verdict
So with all that being said, which platform is better?
Honestly, this answer is simple. Custom GPTs are far superior to Gems. From ease of creation to sharing and discovering creations, ChatGPT surpasses Gemini in every regard. It seems to be that while custom GPTs are a finished and polished product, Gems are still very much a work in progress.
Here is my advice, if you’re a free user and creating a tool for personal use only, I’d recommend experimenting with Gemini. Gems are still a powerful tool when using the Gemini 2.5 models. You also won’t be limited by the sharing aspect. However, if you’re creating a tool you want to share publicly or within a team, I’d recommend using ChatGPT.
Reply