CustomProviders

Many AI providers offer OpenAI-compatible APIs. Instead of maintaining separate SDKs for each, tryAGI.OpenAI includes CustomProviders factory methods that return a pre-configured OpenAiClient pointing to the correct base URL and with the right defaults.

Installation

dotnet add package tryAGI.OpenAI

Provider table

Provider	Factory Method	Default Chat Model
Azure	`CustomProviders.Azure(key, endpoint)`	`gpt-4o-mini`
DeepInfra	`CustomProviders.DeepInfra(key)`	`Qwen/Qwen2.5-72B-Instruct`
DeepSeek	`CustomProviders.DeepSeek(key)`	`deepseek-chat`
Groq	`CustomProviders.Groq(key)`	`llama-3.3-70b-versatile`
XAi	`CustomProviders.XAi(key)`	`grok-3-mini`
Fireworks	`CustomProviders.Fireworks(key)`	`llama-v3p3-70b-instruct`
OpenRouter	`CustomProviders.OpenRouter(key)`	`meta-llama/llama-3.2-3b-instruct:free`
Together	`CustomProviders.Together(key)`	`Llama-3.3-70B-Instruct-Turbo`
Perplexity	`CustomProviders.Perplexity(key)`	`sonar`
SambaNova	`CustomProviders.SambaNova(key)`	`Meta-Llama-3.3-70B-Instruct`
Mistral	`CustomProviders.Mistral(key)`	`mistral-large-latest`
Codestral	`CustomProviders.Codestral(key)`	`codestral-latest`
Cerebras	`CustomProviders.Cerebras(key)`	`llama3.1-70b`
Cohere	`CustomProviders.Cohere(key)`	`command-r-08-2024`
Nebius	`CustomProviders.Nebius(key)`	`Qwen/Qwen2.5-72B-Instruct`
Hyperbolic	`CustomProviders.Hyperbolic(key)`	`Llama-3.3-70B-Instruct`
GitHub Models	`CustomProviders.GitHubModels(token)`	`gpt-4o`
Ollama	`CustomProviders.Ollama()`	`llama3.2`
LM Studio	`CustomProviders.LmStudio()`	(user-selected)

Usage

Basic chat

using OpenAI;
using Microsoft.Extensions.AI;

var client = CustomProviders.DeepSeek(apiKey);
IChatClient chatClient = client.AsChatClient("deepseek-chat");

var response = await chatClient.GetResponseAsync("Hello!");
Console.WriteLine(response);

Streaming

var client = CustomProviders.Groq(apiKey);
IChatClient chatClient = client.AsChatClient("llama-3.3-70b-versatile");

await foreach (var update in chatClient.GetStreamingResponseAsync("Tell me a joke."))
{
    Console.Write(update.Text);
}

Embeddings

var client = CustomProviders.DeepInfra(apiKey);
IEmbeddingGenerator<string, Embedding<float>> generator =
    client.AsEmbeddingGenerator("BAAI/bge-en-icl");

var embeddings = await generator.GenerateAsync(["Hello, world!"]);

Local providers

Ollama and LM Studio require no API key:

// Ollama (default: http://localhost:11434)
var ollamaClient = CustomProviders.Ollama();
IChatClient chat = ollamaClient.AsChatClient("llama3.2");

// LM Studio (default: http://localhost:1234)
var lmStudioClient = CustomProviders.LmStudio();
IChatClient chat2 = lmStudioClient.AsChatClient("your-model-name");

CustomProviders vs dedicated SDKs

Aspect	CustomProviders	Dedicated SDK
Installation	`tryAGI.OpenAI` (one package)	`tryAGI.{Provider}` (per provider)
API coverage	OpenAI-compatible endpoints only	Full provider API
MEAI support	`IChatClient` + `IEmbeddingGenerator`	Provider-specific interfaces
Provider-specific features	Limited to OpenAI compatibility	Full access

Use CustomProviders when you only need chat/embeddings and the provider is OpenAI-compatible. Use a dedicated SDK when you need the provider's full API or unique features.