CLU model Authoring service expiry and prediction service(deployments) expiry

Question

CLU model Authoring service expiry and prediction service(deployments) expiry

Nagireddy SunilBhaskar1 0

Hey team, I'm having few questions related to CLU regarding the model expiry and deployments expiry

I'm having projects in CLU, I got to know that the Azure language studio is going to get expired by February16, 2026. please answer the following questions
Will the models give the responses after the date which was indicated?
We cant train or finetune the models further, is the understanding correct?

2 answers

Your answer

Answer 1

Sridhar M 5,330 Microsoft External Staff Moderator

Hi Nagireddy SunilBhaskar1

you're looking for clarification regarding the expiration of your CLU models and deployments. Here’s the scoop on your questions:

Will the models still respond after the expiration date: After the expiration date, which you mentioned as February 16, 2026, your deployed CLU models will no longer be available for making predictions. Once the deployment has expired, API calls to that model will return an error, and you won’t be able to query the model for responses.
About training and fine-tuning post-expiration: After the indicated expiration date, you won't be able to retrain or fine-tune the models. If you want to continue utilizing your models effectively, it’s advisable to retrain and redeploy them before their expiration date using the latest configuration.

To prevent any service interruptions due to model expiration, you can consider the following steps:

Regularly check the expiration timelines for your models and redeploy them with the latest configurations before they expire.
Familiarize yourself with the processes for model updates and deployment, such as using Azure CLI or PowerShell commands mentioned in the relevant documentation.

For further details on deployments and managing model expirations, you might want to refer to the following resources:

Please 'Upvote'(Thumbs-up) and 'Accept' as answer if the reply was helpful. This will be benefitting other community members who face the same issue.

Nagireddy SunilBhaskar1 0 Reputation points

2026-01-28T01:25:22.84+00:00

Please let me know how can I access the existing projects in Azure CLU in Language studio from Microsoft foundry?
Sridhar M 5,330 Reputation points Microsoft External Staff Moderator

2026-01-28T09:18:35.5466667+00:00

Hi Nagireddy SunilBhaskar1

You cannot directly open or import existing Azure Conversational Language Understanding (CLU) projects from Language Studio into Microsoft Azure AI Foundry. The two tools serve different purposes and do not share project‑level visibility or synchronization.

Product boundary clarification: Azure CLU projects are owned and managed by the Azure Language service and are authored, trained, evaluated, and versioned entirely within Language Studio. Microsoft Azure AI Foundry is an agent orchestration and runtime platform, not a training or labeling studio for CLU‑style NLU models.

Azure AI Foundry is designed to consume language intelligence (including CLU) as an external capability via deployed endpoints. Foundry orchestrates prompts, tools, memory, and routing logic, but it deliberately does not manage CLU intents, utterances, entities, or training datasets.

 Required prerequisite in Language Studio:

To use an existing CLU project in Foundry, you must first deploy the CLU project from Language Studio. This deployment generates a production runtime endpoint tied to a specific project name and deployment name, which is mandatory for downstream consumption.

integrate CLU with a Foundry agent :

In Azure AI Foundry, you integrate CLU by treating it as an HTTP/REST tool (or custom action). The agent sends the user’s utterance to the CLU runtime API and receives a structured JSON response containing intents, confidence scores, and extracted entities.

Foundry does not display CLU projects in its UI, does not allow selection of CLU projects during agent creation, and does not support editing, testing, or retraining CLU models. All CLU lifecycle operations remain exclusively in Language Studio.

The identity used by Azure AI Foundry must have appropriate access (for example, Cognitive Services User/Contributor) to the Language resource hosting the CLU deployment. There is no special Foundry‑specific CLU role—standard Azure RBAC applies.

Azure AI Foundry does not replace Language Studio for CLU scenarios. Language Studio remains the system of record for NLU training, while Foundry operates one layer above, coordinating conversation flow and calling CLU only when intent recognition is required.

Recommended enterprise architecture:

The recommended and supported pattern is: Language Studio (train & deploy CLU) → CLU runtime endpoint → Azure AI Foundry agent tool. This clean separation preserves evaluation workflows, avoids accidental retraining risks, and aligns with Microsoft’s product design boundaries.

existing CLU projects cannot be accessed or imported directly into Azure AI Foundry. Instead, you deploy them in Language Studio and consume them via API calls from Foundry agents. Foundry acts as an orchestrator and caller—not an editor—by design.

Answer 2

The models in the Conversational Language Understanding (CLU) service will not be able to provide responses after the deployment expiration date, which is typically twelve months after the training configuration expiration date. If the Azure Language Studio is set to expire on February 16, 2026, this means that any models deployed using that service will also become unavailable for predictions after that date.

Additionally, once the training configuration expires, you cannot train or fine-tune the models further using that configuration. You would need to use a newer supported training configuration version to submit any training or deployment jobs after the expiration date.

Therefore, your understanding is correct: after the indicated expiration date, the models will not respond, and you cannot train or fine-tune them further until you switch to a supported configuration version.

References:

Share via

CLU model Authoring service expiry and prediction service(deployments) expiry

2 answers

Your answer