Getting error "No connection matching model: gpt-5.1"

Question

Getting error "No connection matching model: gpt-5.1"

Nishant Gupta 0 Microsoft Employee

I am using OpenAI SDK (.NET) for connecting to Azure foundry and creating agents using AzureAssistantClient. I have model deployed in Azure Foundry with gpt 5.1 but when I create thread run and tried to call GetRunAsync to get the status of the run, it fails saying "No connection matching model: gpt-5.1".

With same agent configuration, I am using ChatClient and when invoked CompleteChatAsync, it is able to run successfully

Sridhar M 5,335 Reputation points Microsoft External Staff Moderator

2026-02-26T12:47:32.1733333+00:00
Hi Nishant Gupta

When you use the .NET OpenAI SDK with Azure AI Foundry, the same GPT‑5.1 deployment behaves differently depending on the client. Calls made through ChatClient.CompleteChatAsync work, but when you create an agent using AzureAssistantClient, start a thread, and then poll the run using GetRunAsync, it fails with the error “No connection matching model: gpt‑5.1”. This difference is expected based on how the two execution paths are implemented in Foundry

ChatClient uses the Chat Completions / Responses runtime, which talks directly to your GPT‑5.1 deployment. As long as the model is deployed correctly in Azure Foundry and the endpoint, API version, and key are valid, GPT‑5.1 is fully supported for chat-style invocations. That is why CompleteChatAsync succeeds with the same deployment and credentials

AzureAssistantClient does not execute against deployments directly. Instead, it uses the Foundry Agent Service runtime, where agents, threads, and runs are resolved through internal model connections supported by the agent fabric. During GetRunAsync, the Agent Service tries to bind the run to a compatible agent-side model connection. At present, GPT‑5.1 does not have an agent-compatible connection, so the runtime cannot resolve it and throws “No connection matching model: gpt‑5.1”

This error does not mean that the deployment is missing, misnamed, or misconfigured. It specifically means that the Foundry Agent runtime does not support GPT‑5.1 for threads/runs today. The agent can be created successfully because validation is deferred, but execution fails when the run is executed or polled. This is a platform limitation, not an SDK bug or permission issue

As of now, GPT‑5.1 is supported for Chat Completions, but not supported for Foundry Agent Service (threads/runs). Agent execution currently supports a restricted set of models (for example GPT‑4o / GPT‑4.1 and certain Claude models, depending on region). GPT‑5.x models are not yet wired into the agent execution fabric.

Renaming the deployment, switching between gpt‑5.1 and gpt‑5.1‑chat, changing API versions, or modifying SDK configuration will not resolve this error. There is no supported configuration workaround that enables GPT‑5.1 to run inside Agent threads today

Supported workarounds You have two supported paths:

If you must use GPT‑5.1, continue using ChatClient and manage conversation state yourself (outside Agent threads).

If you must use Agents (threads/runs, tools, orchestration), switch the agent’s model to a supported agent-compatible model (for example GPT‑4o / GPT‑4.1) and keep GPT‑5.1 only for non-agent chat scenarios.

Notes:Microsoft has acknowledged this limitation publicly on Microsoft Learn Q&A. There is no published ETA for GPT‑5.1 support in the Foundry Agent Service at this time. Until that changes, the behavior you are seeing is expected and by design

Reference:

https://github.com/MicrosoftDocs/azure-ai-docs/blob/main/articles/ai-foundry/agents/how-to/connected-agents.md

Foundry Models sold directly by Azure

Troubleshoot Deployment Issues

General Troubleshooting for Azure AI Foundry

Thankyou!
SRILAKSHMI C 14,815 Reputation points Microsoft External Staff Moderator

2026-02-27T06:41:44.9633333+00:00

Hi Nishant Gupta,

Did you get any chance to review the above response. Do let me know if you have any further queries.

Thank you!
Sridhar M 5,335 Reputation points Microsoft External Staff Moderator

2026-03-02T12:50:20.3966667+00:00

Hi Nishant Gupta,

Did you get any chance to review the above response. Do let me know if you have any further queries.

Thank you!
Nishant Gupta 0 Reputation points Microsoft Employee

2026-03-02T13:41:33.39+00:00

Hi @Sridhar M ,

Thank you for your response. Is the behavior w.r.t to GTP 5.1 model changed recently? Becuase util Feb 15, 2026 we were not facing the issue and was able to use GPT 5.1 with AzureAssitantClient.
Nishant Gupta 0 Reputation points Microsoft Employee

2026-03-02T15:44:43.37+00:00

Hi @Sridhar M ,

Just now we tried with GPT 4.0 and GPT 4.1 models. They are also not working. Same issue we are getting

Manas Mohanty 14,660 Microsoft External Staff Moderator

Hi Nishant Gupta

You can toggle to new foundry UI and select GPT 5.1 deployment to get support client code too.

Attached screenshot and code for reference.

User's image

// dotnet add package Azure.AI.Projects --version 1.2.*-*
using Azure.AI.Projects;
using Azure.AI.Projects.OpenAI;
using Azure.Identity;
using OpenAI;
using OpenAI.Responses;

#pragma warning disable OPENAI001

const string projectEndpoint = "https://<foundryname>.services.ai.azure.com/api/projects/foundrytest"; const string agentName = "testagent";
const string agentVersion = "2";

// Connect to your project using the endpoint from your project page
// The AzureCliCredential will use your logged-in Azure CLI identity, make sure to run `az login` first
AIProjectClient projectClient = new(endpoint: new Uri(projectEndpoint), tokenProvider: new DefaultAzureCredential());

ProjectResponsesClient responseClient = projectClient.OpenAI.GetProjectResponsesClient();
// Use the agent to generate a response
ResponseResult response = responseClient.CreateResponse(
    "Hello! Tell me a joke.",
    new ResponseCreationOptions { AgentName = agentName, AgentVersion = agentVersion }
);

Console.WriteLine(response.GetOutputText());

Thank you.

1 answer

Your answer

SRILAKSHMI C 14,815 Reputation points Microsoft External Staff Moderator

2026-02-27T06:41:44.9633333+00:00

Hi Nishant Gupta,

Did you get any chance to review the above response. Do let me know if you have any further queries.

Thank you!
Sridhar M 5,335 Reputation points Microsoft External Staff Moderator

2026-03-02T12:50:20.3966667+00:00

Hi Nishant Gupta,

Did you get any chance to review the above response. Do let me know if you have any further queries.

Thank you!
Nishant Gupta 0 Reputation points Microsoft Employee

2026-03-02T13:41:33.39+00:00

Hi @Sridhar M ,

Thank you for your response. Is the behavior w.r.t to GTP 5.1 model changed recently? Becuase util Feb 15, 2026 we were not facing the issue and was able to use GPT 5.1 with AzureAssitantClient.
Nishant Gupta 0 Reputation points Microsoft Employee

2026-03-02T15:44:43.37+00:00

Hi @Sridhar M ,

Just now we tried with GPT 4.0 and GPT 4.1 models. They are also not working. Same issue we are getting
Manas Mohanty 14,660 Reputation points Microsoft External Staff Moderator

2026-03-06T02:01:59.49+00:00

Hi Nishant Gupta

You can toggle to new foundry UI and select GPT 5.1 deployment to get support client code too.

Attached screenshot and code for reference.

// dotnet add package Azure.AI.Projects --version 1.2.*-* using Azure.AI.Projects; using Azure.AI.Projects.OpenAI; using Azure.Identity; using OpenAI; using OpenAI.Responses; #pragma warning disable OPENAI001 const string projectEndpoint = "https://<foundryname>.services.ai.azure.com/api/projects/foundrytest"; const string agentName = "testagent"; const string agentVersion = "2"; // Connect to your project using the endpoint from your project page // The AzureCliCredential will use your logged-in Azure CLI identity, make sure to run `az login` first AIProjectClient projectClient = new(endpoint: new Uri(projectEndpoint), tokenProvider: new DefaultAzureCredential()); ProjectResponsesClient responseClient = projectClient.OpenAI.GetProjectResponsesClient(); // Use the agent to generate a response ResponseResult response = responseClient.CreateResponse( "Hello! Tell me a joke.", new ResponseCreationOptions { AgentName = agentName, AgentVersion = agentVersion } ); Console.WriteLine(response.GetOutputText());

Thank you.

Answer 1

Hi Nishant Gupta

I am trying to replicate this issue to be in same page.

It seems you are using "AzureAssistantClient" to create Agents and facing some issue while using "GetRunAsync" to get the status of the runs.

The codes mentioned in official documentation for Foundry classic uses PersistentAgent client though.(GPT 5.1 not supported)

https://learn.microsoft.com/en-us/azure/foundry-classic/agents/quickstart?pivots=programming-language-csharp

Also, attached code for new foundry which uses "projectClient" (GPT 5.1 supported)

https://learn.microsoft.com/en-us/azure/foundry/quickstarts/get-started-code?tabs=csharp

Please share your observation with official documentation usage.

Have shared my connect details via private message to connect post minimal replication with respect to above official documentation to find any recent regression issues.

Thank you for your inputs.

Share via

Getting error "No connection matching model: gpt-5.1"

1 answer

Your answer