
UPDATED: Jan 2025
When Google unveiled Gems I was really excited. For me Open AI’s Custom GPTs was essential functionality Google Gemini was lacking. The addition of a Customisable AI Chatbot to Gemini would go a long way to bridging the gap in terms of user experience.
However when we first tested Gems back in August there were some glaring holes in it's functionality mainly the ability to upload files to a knowledge base. However I think the powers that be must have been reading this blog as they have since made a number of improvement including the ability to add knowledge based files to Gems!
Gems vs Custom GPT’s
Feature | Gems | Custom GPT |
Ability to give instructions | Yes | Yes |
Ability to save and come back to the Custom Gem/GPTÂ | Good | Good |
Add knowledge base files | Yes | Yes |
Able to share GPT/Gem with Colleagues | No | Yes |
Able to share GPT/Gem with via URL only | No | Yes |
Able to share GPT/Gem Publicly | No | Yes |
Gem/GPT Store | Google only Gems | GPT's from Open AI & any creator |
Security | Most Secure | Security Risk |
Â
Knowledge Base Integration:Â By fixing this Google has done a lot to bridge the gap to Custom GPT's. The main differences now revolve around sharing and public access. However this could be intentional due to the data security risks associated with CustomGPT's particularly for corporate users.
No Sharing Capabilities:Â Unlike Custom GPTs, (which can be private, shared with your team, shared via a private URL or published publicly) Gems are private. Whilst this limits their utility, especially for teams or businesses looking to implement AI solutions across their operations, it does make them more secure.
No Public Accessibility: Another major drawback is the absence of a public platform for Gems. At present you can only access other Gems that have been created by Google. Custom GPTs can be published on the GPT store, allowing anyone to access and benefit from them. However, I should add that the ability for anyone to publish a Custom GPT on the GPT store is a double edge sword. It is a huge benefit for publishers but means that the quality of Custom GPT’s can vary wildly by publisher and you never really know what you are getting. So in this regard I can understand why Google has refrained from launching a full GPT store equivalent.
Â
Why Custom GPTs Still Lead the Market
In contrast to Gems, Custom GPTs remain the go-to for those looking for accessible, customisable, and shareable AI tools. Whether you’re looking to automate customer service, streamline content creation, or develop industry-specific AI solutions, Custom GPTs offer a robust platform with the flexibility and power needed to get the job done.
Google has taken a big step with the addition of a knowledge base for Gems and as such the variety of use cases they can be applied to has increased.
If you’re serious about leveraging AI in your business or personal projects, Custom GPTs are still your best bet but just be aware that if you share them publicly your data in the knowledge base is publicly accessible. Also when using GPT's on the GPT store be aware that your data can be passed to a third party system too.
I am sure that further improvements will be made to Gem's in the coming months, so keep your eyes peeled!

For now, ChatGPT’s Custom GPTs remain the most accessible and user-friendly general purpose AI tools but things tend to change very quickly in the world of AI.
Â