`StreamingChatCompletionUpdate.FinishReason` is always null

[mgesing]

Service

OpenAI

Describe the bug

For a CompleteChatStreamingAsync request the property StreamingChatCompletionUpdate.FinishReason is always null. Stop is expected when the response has been transmitted in full or length when the maximum number of tokens per request has been reached.

In comparison, ChatCompletion.FinishReason is set correctly when using CompleteChatAsync.

Steps to reproduce

Set ChatCompletionOptions.MaxOutputTokenCount to a value that is less than the tokens required for the response and make a request using CompleteChatStreamingAsync.

Code snippets

OS

Windows 10

.NET version

.NET 9.0

Library version

OpenAI 2.1.0

[joseharriaga]

Thank you for reaching out, @mgesing ! I'm not able to repro this issue with the latest version of the library.

Here is the code that I tried:

ChatClient client = new(model: "gpt-4o", apiKey: Environment.GetEnvironmentVariable("OPENAI_API_KEY"));

ChatCompletionOptions options = new()
{
    MaxOutputTokenCount = 2,
};

AsyncCollectionResult<StreamingChatCompletionUpdate> completionUpdates = client.CompleteChatStreamingAsync(
    ["Say 'this is a test.'"],
    options);

Console.Write($"[ASSISTANT]: ");
await foreach (StreamingChatCompletionUpdate completionUpdate in completionUpdates)
{
    if (completionUpdate.FinishReason == ChatFinishReason.Length)
    {
        Console.WriteLine();
        Console.WriteLine("No repro.");
        break;
    }

    if (completionUpdate.ContentUpdate.Count > 0)
    {
        Console.Write(completionUpdate.ContentUpdate[0].Text);
    }
}