A few months ago the OpenAI .NET library was released as a beta. It has now reached version 2.0.0 and the time has come to leave beta and, with a few amendments enter production readiness.
As a refresher, the library provides convenient access to the OpenAI REST API from .NET applications so that developers can build AI driven applications. That access breaks down to :
Support for the entire OpenAI API, including Assistants v2 and Chat Completions
Support for GPT-4o, OpenAI’s latest flagship model
Extensibility to enable the community to build libraries on top
Sync and async APIs for ease of use and efficiency
Access to streaming completions via IAsyncEnumerable<T>
.NET Standard 2.0 compatibility: written in C#, supports all .NET variants that implement .NET Standard 2.0, ensuring compatibility with the latest .NET platforms.
You can very easily add it to your project through NuGet:
dotnet add package OpenAI
With that in place and after obtaining an API key, you can enrich you application with the following functionalities:
chat completions with streaming
chat completions with tools and function calling
chat completions with structured outputs
generate text embeddings
generate images
transcribe audio
use assistants with retrieval augmented generation (RAG)
use assistants with streaming and vision
To put it into practice, here's a sample that summons various clients; embedding, image and audio.
The new version 2.0.0 implements a few breaking changes:
Renamed `OpenAIClientOptions`'s `ApplicationId` to `UserAgentApplicationId` (commit_hash)
Replaced the deprecated `ChatFunctionChoice(ChatFunction)` constructor with `CreateNamedChoice(string functionName)`
Renamed `FileClient` to `OpenAIFileClient` and the corresponding `GetFileClient()` method in `OpenAIClient` to `GetOpenAIFileClient()`
Renamed `ModelClient` to `OpenAIModelClient` and the corresponding `GetModelClient()` method in `OpenAIClient` to `GetOpenAIModelClient()`
So if you've been using the beta and want to upgrade, take those into consideration.
There's also another Microsoft AI SDK, that of Semantic Kernel which serves a similar purpose in accessing LLMs from your code. The main difference is that with Semantic you get access to multiple models, and not just OpenAI’s, plus you can choose between the C#, Java, Python and Javascript versions. Explore both options to see which fits your needs.
The TypeScript team at Microsoft has released details of ongoing work they've begun on - a native port of the TypeScript compiler and tools with Go as the language of choice for the project.
The great hope for AI is that it can solve difficult problems, reducing costs and making solutions widely accessible. Aarvark is a weather forecasting system that can be run on a single desktop comput [ ... ]
