Expanding Accessibility of Custom GPTs: A Call to OpenAI and Its Community

[iankimaru]

As an enthusiastic developer and user of OpenAI’s revolutionary GPT-4 technology, I’ve ventured into the realm of creating a custom GPT model. However, a significant challenge has emerged: the limited accessibility of these custom models for the majority of ChatGPT users, particularly those using the free version.

The Challenge with Current Accessibility The current framework requires users to have a subscription to access GPT-4, including custom models shared by developers like myself. This restriction creates a barrier, preventing a vast majority of the ChatGPT user base from experiencing and benefiting from these innovative custom models.

Impact on Developers and Revenue Sharing As a developer, this limitation hinders the potential for benefiting from OpenAI’s promising revenue-sharing model. The model’s success and viability are contingent upon widespread accessibility and usage. Without the ability to reach the broader, free user base, the opportunity to generate revenue through popular custom GPTs remains largely untapped.

The Case for Broader Accessibility Making custom GPT models accessible to free users is not just beneficial for developers seeking revenue-sharing opportunities. It also aligns with the broader goal of fostering innovation and adaptation within the AI community. By allowing free users to access these custom models, OpenAI can encourage more people to explore and engage with GPT technology, potentially leading to greater acceptance and integration of AI in various fields.

Looking Forward In light of these considerations, I look forward to a future where OpenAI expands accessibility to custom GPT models for free users. Such a move would not only empower developers like myself but also enrich the user experience for the entire ChatGPT community. It represents a step towards a more inclusive, innovative, and collaborative AI ecosystem.

OpenAI has always been at the forefront of AI innovation, and making custom GPTs accessible to a broader audience would further solidify its position as a leader in the field. As both a developer and an advocate for AI accessibility, I urge OpenAI to consider this proposal for the benefit of the entire community.

[bill.french]

What’s the downside of building a custom GPT as an Assistant and then just calling it? Cost? Performance?

Here’s a trick I use with clients who have created assistants and then want to ise that as the basis for a chat completion -

• Armed with the assistant ID, you can instantiate the assistant (which is only 80ms, and at no AI cost to the account).
• You can read the data inside the assistant. It will contain the instructions set by the creator and the model they chose.
• Extract the model and the instructions and then build a prompt and use the model directly with a chat completion call.

This makes it possible to slash costs and improve throughput significantly while repurposing the assistant as a callable custom GPT.

Thoughts?