ChatGPT 4o Image Generation via API

[Jim]

I see that today OpenAI have released access to their latest image gen model via the API.

I’ve created a Pickaxe using GPT-4o, given it my API key (and my keys are organisation verified) and attached it to the Studio where my test Pickaxe is. I can see that the API key gets used, but an image is not returned.

Is there something else that needs to be put into place for this to work from the Pickaxe end? I notice that the model is called ‘gpt-image-1’ so presumably it’s something that we’ll have to invoke differently rather than directly via GPT-4o in Pickaxe?

[ab2308]

@Jim correct it is a different model and will need to be enabled / integrated by Pickaxe

[ihmunro]

you could also call an action to run a make.com scenario to generate the image

[data_interactive]

can you please share the make flow. I’m new to this.

[Jim]

You’ll have to create a JSON module and map the API requirements in, and then pass those through an API call to OpenAI. That’s as far as I got - I can get it to create an image, but can’t seem to get the webhook to pass it back to the Pickaxe.

My blueprint is attached, it needs a little work. You can see the last two modules are trying to pass the image back but I’ve been unsuccessful.

[ab2308]

Hi @Jim I think the response body might be the problem. I can’t see the Pickaxe action you are using but it normally expects a simple JSON response.

If successfull you should get a link to download the image: