.NET: [Bug]: Cannot set ReasoningEffort on gpt-5.6-luna

Description

I created a HarnessAgent backed by a gpt-5.6-luna deployment on Foundry.

After I set the ReasoningOptions, I started getting 400 responses from the Foundry service.

Code Sample

I use the ChatClient from the Aspire Azure OpenAI integration:

var azureOpenAIDeploymentName = builder.Configuration["AzureOpenAI:DeploymentName"];

builder.AddAzureOpenAIClient(connectionName: "openai")
       .AddChatClient(azureOpenAIDeploymentName); // model deployment name from Azure AI Foundry

return sp.GetRequiredService<IChatClient>().AsHarnessAgent(new HarnessAgentOptions
        {
            Name = name,
            HarnessInstructions = Instructions,
            ChatOptions = new ChatOptions { Tools = tools, Reasoning = new() { Effort = ReasoningEffort.Medium, Output = ReasoningOutput.Summary} },
            AIContextProviders = contextProviders,
            DisableWebSearch = true,
            OpenTelemetrySourceName = GenAiSourceNames.Agents,
        });

Error Messages / Stack Traces

HTTP 400 (invalid_request_error: )

Parameter: reasoning_effort

Function tools with reasoning_effort are not supported for gpt-5.6-luna in /v1/chat/completions. To use function tools, use /v1/responses or set reasoning_effort to 'none'.

Package Versions

Microsoft.Agents.AI.Harness: 1.21.0, Aspire.Azure.AI.OpenAI: 13.5.4-preview.1.26464.4

.NET Version

.NET 10.0

Additional Context

The url that Agent Framework calls is:

https://.openai.azure.com/openai/deployments/gpt-5.6-luna/chat/completions?*

It is unclear why the (Azure) OpenAI ChatClient does not use the Responses API instead of the (legacy) Chat Completions API.