[FEATURE REQ] Expose API responses object model's custom deserialization

[origine999]

Library name

Azure.AI.OpenAI

Please describe the feature.

Azure.AI.OpenAI has custom code for deserializing the REST API's responses (for instance completions, chat completions).

For instance: internal static Completions DeserializeCompletions(JsonElement element, ModelReaderWriterOptions options = null)

The problem is that if I get a REST API response from another mean (let's say through a proxy LLM API), it's not possible to leverage this library's object model for completions.

Solution would be to expose this custom deserialization logic to allow converting a string response to Azure.AI.OpenAI's object model.

Note: another benefit of this is let's say you have a CSV storing thousands of OpenAI API responses, you could load it up in a C# program and rehydrate those using the object model.

[github-actions]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @jpalvarezl @trrwilson.

[origine999]

Closing, missed that we can use System.ClientModel: https://learn.microsoft.com/en-us/dotnet/api/overview/azure/system.clientmodel-readme?view=azure-dotnet#simple-modelreaderwriter-usage.