# OpenAI

### Can I use my ChatGPT Plus/Pro subscription for my OpenAI connected model?

No, ChatGPT and the OpenAI API Platform are different products. OpenAI Platform credits are separate from ChatGPT Plus and Pro subscriptions.

If you want to purchase OpenAI API credits from OpenAI rather than using Relay.app's built-in [AI credits](https://docs.relay.app/ai/ai-credits), you'll need to create an OpenAI Platform account and add a payment method. Read more about how to purchase them from OpenAI [here](https://www.relay.app/blog/how-to-buy-credits-for-the-openai-platform-and-api).

### Can I use my OpenAI API key to run OpenAI steps?

Yes. On an AI step, select OpenAI as the model provider, connect your API key, and choose a model from the list of available models for your account.

If this is your preferred model, you can set it as the default model for yourself and/or your team in [settings](https://run.relay.app/dashboard#/settings/account/ai).

If your team is using the same API key, you can also share the connection with Relay.app teammates in the [apps page](https://run.relay.app/dashboard#/settings/apps?app=openai).

<figure><img src="https://647519108-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXkKoEy3xMIYaIwd4mRAB%2Fuploads%2FaljRmdmKcAfIYu2zspoO%2Fimage.png?alt=media&#x26;token=74188491-0412-4532-a028-a1ec8163233e" alt="Screenshot of OpenAI model options"><figcaption></figcaption></figure>

### Can I use my OpenAI fine-tuned model in Relay.app?

Yes, you can use a fine-tuned OpenAI model in Relay.app. Connect your OpenAI Platform account, choose to use your own credentials for an AI step, and select your fine-tuned model.

There are a few caveats to using fine-tuned models, as they can reduce performance and require extensive training data. Read more in [this blog post](https://www.relay.app/blog/assistants_finetunedmodels_customgpts).

### Can I access my OpenAI Assistant with Relay.app?

No. Assistant support is not on our roadmap, as OpenAI has announced its [deprecation](https://platform.openai.com/docs/guides/responses-vs-chat-completions#assistants). As an alternative, we suggest using a regular AI prompt step within Relay.app, which allows connecting knowledge sources using the Relay.app [knowledge](https://docs.relay.app/ai/knowledge "mention") feature (with support for knowledge bases you may have already uploaded to OpenAI coming soon) and a custom system prompt, which can be added within Relay.app as well:\
\
![](https://647519108-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FXkKoEy3xMIYaIwd4mRAB%2Fuploads%2Fgit-blob-a72ae0e8a89880d1e98c609051ca3a029c203173%2Fimage.png?alt=media)\
\
If using an OpenAI Assistant is a key use case for your needs and the workaround above isn't sufficient, please let us know at <support@relay.app>.

### Can I access my OpenAI Custom GPTs with Relay.app?

Unfortunately, custom GPTs are not exposed in the OpenAI API platform. Without API support, Relay.app (and other automation products) are unable to access custom GPTs. To achieve the same goal, you can use the Relay.app [knowledge](https://docs.relay.app/ai/knowledge "mention") feature and input the same data as you used in your custom GPT.

### Does Relay.app or OpenAI use my data to train their models?

In short, no. Relay.app is not a model provider, so we don't use your data to train models. By default, OpenAI does **not** train models on inputs and outputs through their API. [Learn more](https://platform.openai.com/docs/concepts)
