Migrating from Legacy OpenAI .NET SDK: What Replaces CompletionRequest in Version 2.1.0?

[Ethan99]

I’m having trouble after upgrading my OpenAI .NET SDK from version 1.11.0 to 2.1.0. The classes I was using before don’t seem to exist anymore.

In my old implementation, I was using these components that are now missing:

• CompletionRequest class
• Completions.CreateCompletionAsync method

Here’s my original working code from the older SDK version:

using OpenAI_API;
using OpenAI_API.Completions;

public class AiContentGenerator : IContentGenerator
{
    private readonly OpenAIAPI _apiClient;
    private readonly IUserValidator _validator;

    public AiContentGenerator(string key, IUserValidator validator)
    {
        _apiClient = new OpenAIAPI(key);
        _validator = validator;
    }

    public async Task<ContentResult> GenerateContentAsync(ContentRequest input)
    {
        var promptRequest = new CompletionRequest
        {
            Prompt = input.UserPrompt,
            Model = input.SelectedModel,
            MaxTokens = input.TokenLimit,
            Temperature = input.Creativity
        };

        var result = await _apiClient.Completions.CreateCompletionAsync(promptRequest);
        var cleanedOutput = result.ToString().Replace("\n", "").Replace("\r", "").Trim('"', ' ');

        return new ContentResult(cleanedOutput, 0, true);
    }
}

After updating to 2.1.0, I tried modifying the code like this:

using System.Threading.Tasks;
using OpenAI;

public class AiContentGenerator : IContentGenerator
{
    private readonly OpenAIClient _openAiClient;
    private readonly IUserValidator _validator;

    public AiContentGenerator(string key, IUserValidator validator)
    {
        _openAiClient = new OpenAIClient(key);
        _validator = validator;
    }

    public async Task<ContentResult> GenerateContentAsync(ContentRequest input)
    {
        var promptRequest = new CompletionRequest
        {
            Prompt = input.UserPrompt,
            Model = input.SelectedModel,
            MaxTokens = input.TokenLimit,
            Temperature = input.Creativity
        };

        var result = await _openAiClient.Completions.CreateCompletionAsync(promptRequest);
        var cleanedOutput = result.ToString().Replace("\n", "").Replace("\r", "").Trim('"', ' ');

        return new ContentResult(cleanedOutput, 0, true);
    }
}

But this doesn’t work because:

1. CompletionRequest class doesn’t exist
2. Completions property is not available on OpenAIClient

My Setup:

• Framework: .NET 9.0
• Package: OpenAI version 2.1.0
• Also using: Newtonsoft.Json version 13.0.3 and Rystem.OpenAi version 4.0.5

What I need help with:

1. How do I make completion requests in OpenAI .NET SDK version 2.1.0?
2. What classes or methods should I use instead of the old CompletionRequest and Completions.CreateCompletionAsync?
3. Is there a different way to handle text completions in this newer version?

I’d really appreciate any code examples or guidance on how to properly migrate this functionality. Thanks!

[Alex_Brave]

Hey! Hit this same issue last month. The completion API’s deprecated now - you need ChatCompletions instead. Try await _openAiClient.GetChatClient("gpt-3.5-turbo").CompleteChatAsync() and pass your prompt as a user message. Different structure but works great once you’ve got it set up.