Batch runners provide helpers for OpenAI and Azure OpenAI batch jobs.

• OpenAI: dotnet add package AgentFrameworkToolkit.OpenAI
• Azure OpenAI: dotnet add package AgentFrameworkToolkit.AzureOpenAI

OpenAIBatchRunner batchRunner = new("<apiKey>");

ChatBatchRun run = await batchRunner.RunChatBatchAsync(
    new ChatBatchOptions
    {
        Model = "gpt-5-mini"
    },
    [
        ChatBatchRequest.Create("Summarize this text.")
    ]);

OpenAI batch runners support:

• Chat batches with RunChatBatchAsync
• Structured-output chat batches with RunChatBatchAsync<T>
• Existing chat batch lookup with GetChatBatchAsync
• Existing structured chat batch lookup with GetChatBatchAsync<T>
• Embedding batches with RunEmbeddingBatchAsync
• Existing embedding batch lookup with GetEmbeddingBatchAsync

AzureOpenAIBatchRunner batchRunner = new("<endpoint>", "<apiKey>");

ChatBatchRun run = await batchRunner.RunChatBatchAsync(
    new ChatBatchOptions
    {
        Model = "<deployment-name>"
    },
    [
        ChatBatchRequest.Create("Summarize this text.")
    ]);

Azure OpenAI batch runners support:

• Chat batches with RunChatBatchAsync
• Structured-output chat batches with RunChatBatchAsync<T>
• Existing chat batch lookup with GetChatBatchAsync
• Existing structured chat batch lookup with GetChatBatchAsync<T>

AzureOpenAIBatchRunner withApiKey = new("<endpoint>", "<apiKey>");
AzureOpenAIBatchRunner withRbac = new("<endpoint>", new AzureCliCredential());
OpenAIBatchRunner withConnection = new(new OpenAIConnection { ApiKey = "<apiKey>" });

• In Azure OpenAI, Model is your deployment name.
• Batch runners reuse the same connection objects as the corresponding provider factories.