Integrate GPT-4 into Model's List

As we continuously strive to innovate and stay ahead in the competitive landscape of AI-driven solutions, I am writing to request the integration of OpenAI’s GPT-4 model into our existing list of AI models for Open AI integrations.

The OpenAI integration has many features. Most are meant to be as simple as possible and preconfigured, and a couple allow you to define which model you would like to use. The Complete Prompt feature, for instance, allows you to configure model, temperature, maximum length, frequency penalty. Glide will probably not allow this for all features, it would make them too complicated to use for most people.

When you can configure the model, check OpenAI’s documentation to see which model might suit your use case best, and test with different models to see if you get a satisfactory output.

Perhaps GPT-4 is not yet offered because it is or was in limited beta.

1 Like

I appreciate your explanation of the OpenAI integration features and their varying levels of complexity. While it’s essential to maintain a balance between simplicity and advanced options for Glide users, I believe there could be an opportunity to further enhance the AI capabilities by including GPT-4 in the models list for those who have access.

As GPT-4 is a significant advancement in AI, offering state-of-the-art performance, I propose that Glide considers adding it as an optional model for users who are eligible for access. This could potentially benefit Glide by:

  1. Attracting users seeking the latest AI advancements: By offering GPT-4 as an option, Glide can appeal to users who value cutting-edge technologies and capabilities.
  2. Encouraging AI experimentation: Users with access to GPT-4 can explore its features and compare its performance with other models, potentially discovering new use cases and applications that Glide can support.
  3. Strengthening Glide’s position as an AI leader: Incorporating GPT-4 in the models list would demonstrate Glide’s commitment to staying current with AI advancements, enhancing its reputation as a leading AI platform.

I hope that Glide’s team consider this proposal and assess the potential benefits of adding GPT-4 to the models list. It could be an excellent opportunity to further enhance Glide’s AI offerings and contribute to the progress of AI technology.

1 Like

I’ve already moved on from Glide because this feature is missing. It’s too bad, because overall I liked it. But I can get much better features (including GTP 4 integration) elsewhere.

You will be missed :sob:

What tool are you using now?

I went to word press

GPT-4 is already released. What’s a problem to just add the model gpt-4 to the models list?

Look for chatgpt when search actions//columns and use the gpt-4 model override. Note you need to have paid access as well as elevated access to their api (I still think that’s in beta on their end afaik)


I beg to differ, the problem doesn’t seem to be resolved. The chat function isn’t equivalent to the “prompt completion” function, and ideally, the gpt-4 model should be integrated within the “complete prompt” for optimal outcomes. It’s peculiar that in the “chatGPT” function, the gpt-4 model is employed, but this isn’t the case in the Complete prompt.

Point taken.

OpenAI explicitly made different entrypoint’s to these API’s with different performance (one shot vs multi example preference for example) and interfaces (simple text in vs a structured role based set of messages) so we tried to align with that more assuming they’re going to iterate on the features on the chat interface. If you use the Send message to ChatGPT (with message history) as an action instead of column you’ll see why they’re different. The column version (without history) simplifies their already opinionated way of getting data to them.

I’ll add your feedback to the team as far as unifying the case you made, Thanks.

1 Like

Why is it so hard to add a “gpt-4” in model list? Is there any other way beside sending message to ChatGPT? Your app already has nodes that support model choice. Sounds like making difficulties where they can be avoided.

This topic was automatically closed 24 hours after the last reply. New replies are no longer allowed.