BUG: images/edits API is giving confusing responses

API Bugs
1

I was testing the API with hazel-edit-6 model which was working until 09/12 however since December 11th the api has been responding with this error:

{\n \“error\”: {\n \“message\”: \“The model ‘hazel-edit-6’ does not exist.\”,\n \“type\”: \“invalid_request_error\”,\n \“param\”: null,\n \“code\”: \“model_not_found\”\n }\n}

What’s confusing because when I give a model that I know doesn’t exist they responds with:

{ “error”: { “message”: “Invalid value: ‘dall-e’ Supported values are: ‘dall-e-2’, ‘gpt-image-1’, ‘gpt-image-1-mini’, ‘gpt-image-1.5’, ‘gpt-image-1.5-mini’, ‘hazel-gen-2’, ‘hazel-gen-3’, ‘hazel-gen-4’, ‘hazel-edit-2’, ‘hazel-edit-4’, ‘hazel-edit-5’, ‘hazel-edit-6’, ‘hazel-edit-7’, and ‘hazel-small-2’.”, “type”: “invalid_request_error”, “param”: null, “code”: “invalid_value” }

}Why this is happening?

2

List models via the models endpoint API call.

Those are the ones your organization should have access to, with the only officially released edits endpoint models being dall-e-2 and gpt-image-1.

One can speculate that other model names are “validation” names so that early testers can pass them to the endpoint without rejection, and others able to make use of any unpublished model name is a “leak”, which has been corrected.

You might have to wait for a holiday gift to see if a better pixel-accurate mask edit can be delivered…

Plus, using models that aren’t in a price list is literally a gamble.

1 Like
3

Thanks. I suppose that’s the reality.

That’s very sad because I used Hazel-6 for two days and the results were amazing! I searched for a release date, but I didn’t find anything.