[FEATURE REQ] OpenAI: Make output of ImageGenerationCall consistent with ImageGeneration response

[ericstj]

Describe the feature or improvement you are requesting

When generating an image through the responses API very little information is available with the image.
https://platform.openai.com/docs/api-reference/responses/object

Contrast that to the image generation API where much is available:
https://platform.openai.com/docs/api-reference/images/object

output_format in particular is most important since it determines how the data can be interpreted - what it's mime-type or file extensions should be. I can see that this information is provided by the response, but it's undocumented as being avaialble.

While some of this information is documented as supported on the request, it might also be not be provided and left to whatever the default is for the model. There is no way to discover this default as a client. As a result we end up hardcoding things that might change.

Additional context

See discussion in #702 (comment)

[jsquire]

Hi @ericstj. Thanks for reaching out and we regret that you're experiencing difficulties. As you noted, the properties do not appear on the platform documentation nor the REST API specification, which means that they're not current contract. Once they are available in the spec, we'll be able to add them to the client. I'll route this to the appropriate upstream location to request that the folks who manage the spec investigate.