I’ll walk you through some of the few avenues you have for self-help before OpenAI would have to repair the organization, via “help” messages on the platform site.
Ensure you have the correct hierarchy of organization selected at the top-left of the platform site page
Create a new project as a container for API keys. Don’t set any model or endpoint prohibitions on it, nor limits. Ensure it is then selected with that organization->project selector at upper-left
Try the platform site for making your images: https://platform.openai.com/playground/images. This should run independent of any API keys. Pick “low” quality for cheap testing. It only uses gpt-image-1 over the generate or edits API.
Then within the project, create your new API key, and copy it out for use.
Ensure your application is actually using the new API key. Here, for example, is a Python environment test, also ensuring you don’t have conflicting old values of org or project which would be revealed in its printout:
from openai import OpenAI
client = OpenAI()
response = client.images.generate(
model="gpt-image-1",
prompt="Photo: A kitten who is an astronaut on a space walk.",
size="1024x1024",
quality="low",
background="opaque",
output_format="png",
)
image_base64 = response.data[0].b64_json
print("Credentials used:\n"
f"{client.api_key}\n{client.organization}\n{client.project}")
# your code here, or save from base64 response
import base64
with open("testsave.png", "wb") as f:
f.write(base64.b64decode(image_base64))
