Step-by-Step Guide to Creating a Custom GPT
Last updated
Last updated
Step 1 - Access the GPT Builder
Log in to your OpenAI account and go to the .
In the left-hand sidebar, click on the "Explore" option.
On the Explore page, you will see a section for "Create a GPT". Click on the "Create" button.
In the "Create" section, you will see a message box where you can provide instructions for your Custom GPT. You can either configure the Custom manually or via a direct conversation with ChatGPT (Tab Create / Tab Configure)
Name
Name your GPT. It will appear under the CustomGPT Profile Picture
Example
Description
Describe the Role of the Assistant
Example
Instructions
Describe the purpose and capabilities of your Custom GPT. For example, you could create a "Product Manager GPT" with the following instructions:
Example:
Conversation Starters
Help the user start the conversation with the chatbot and offer 3 - 5 examples of possible questions, that kick-off the dialogue. Look for t
If you upload files under Knowledge, conversations with your GPT may include file contents. Files can be downloaded when Code Interpreter is enabled
In the "Configure" section, you can upload files or documents that you want your Custom GPT to reference when generating responses.
This could include product management frameworks, user personas, market research reports, or any other relevant information.
To demonstrate, let's say you want to upload a product manager job description. You can do so by clicking the "Upload" button in the "Knowledge" section and selecting the file.
Example Prompt
In the "Configure" section, you can enable or disable various capabilities for your Custom GPT, such as web browsing, image generation, or code execution.
For a Product Manager GPT, you might want to enable web browsing to allow the AI to research relevant topics, and disable image generation, as it may not be a core requirement.
According to the search results, OpenAI Custom GPTs support a variety of file types for knowledge sources and file uploads:
Knowledge sources can include many file types, such as PDF, text, CSV, and other data formats[1][2]. These files are used to extend the GPT's knowledge with domain-specific information.
For file uploads, all common file extensions for text files, spreadsheets, presentations, and documents are supported[3]. The maximum file size is 512MB per file, with a 2M token limit for text and document files (this limitation does not apply to spreadsheets)[3].
Images can also be uploaded, with a size limit of 20MB per image[3]. However, images embedded in documents or presentations are not yet supported[3].
Files uploaded as knowledge to a Custom GPT are retained indefinitely, while files used for other purposes (e.g., Advanced Data Analysis) are retained for a duration that varies based on the user's plan[3].
In summary, OpenAI Custom GPTs support a wide range of file types, including text files, spreadsheets, presentations, documents, and images, which can be used to extend the GPT's knowledge and capabilities for specific use cases.
References
[1] OpenAI Custom GPTs: What You Need to Worry About https://www.paloaltonetworks.com/blog/prisma-cloud/openai-custom-gpts-security/
[2] OpenAI GPT-3 API: Which file formats can be used for fine-tuning? https://stackoverflow.com/questions/75559672/openai-gpt-3-api-which-file-formats-can-be-used-for-fine-tuning
[3] File uploads FAQ - OpenAI Help Center https://help.openai.com/en/articles/8555545-file-uploads-faq
[4] Introducing GPTs - OpenAI https://openai.com/blog/introducing-gpts
Based on the search results, the key file size and upload limits for OpenAI Custom GPTs are:
Maximum number of files per GPT: 20 files[3]
Maximum file size per file: 512 MB[2]
Maximum total file size across all files uploaded to a GPT: 100 GB per organization[2]
Maximum token count for text/document files (excluding spreadsheets): 2 million tokens per file[2]
Maximum image file size: 20 MB per image[2]
The search results also indicate that images embedded within documents or presentations are not yet supported[2].
Additionally, the files uploaded as knowledge sources to a Custom GPT are retained indefinitely, while files used for other purposes (e.g., Advanced Data Analysis) are retained for a duration that varies based on the user's plan[2].
So in summary, the key limits are 20 files per GPT, 512 MB per file, and a 100 GB total per organization, with some additional restrictions on token counts and image sizes.
References
[1] Number of files you can upload on a GPT - OpenAI Developer Forum https://community.openai.com/t/number-of-files-you-can-upload-on-a-gpt/523616
[2] File uploads FAQ - OpenAI Help Center https://help.openai.com/en/articles/8555545-file-uploads-faq
[3] GPTs knowledge capacity limits - OpenAI Developer Forum https://community.openai.com/t/gpts-knowledge-capacity-limits/492955
[4] What is the max amount of data an Azure OpenAI Gpt-4 model can ... https://learn.microsoft.com/en-us/answers/questions/1600250/what-is-the-max-amount-of-data-an-azure-openai-gpt
Once you're satisfied with your Custom GPT's configuration, click the "Save" button in the top-right corner.
You can then choose to make your Custom GPT available only to yourself, your organization, or publicly to the entire OpenAI community.
If you decide to publish your Custom GPT publicly, you can share the link with others or embed it on a website.