GPT-5.1 Lands in Copilot Studio: What It Means for Custom Agents

Following yesterday's launch, GPT-5.1 models came to Copilot Studio immediately. The list of available models as the primary reasoning model for custom agents continues to grow impressively.

But for teams evaluating which available model might be the best fit for a particular purpose, this list still pales in comparison to the fuller selection available in Azure AI Foundry. However, you can plug these models into AI Prompts used throughout Power Platform if you have a desire to use something more exotic inside the confines of Copilot Studio.

Increasingly though, model and platform selection can come down to different criteria than just which is the latest or smartest. For example, I've recently been exploring the fairly unique citation management capabilities that Claude models offer with their API implementation. Claude citations on custom content built from meeting transcripts might be a topic for an upcoming post or video. These types of features are not ones that grab headlines but are really important to being able to easily achieve a suitable business outcome.

There's lots of reasons to be really positive about Copilot Studio so quickly following OpenAI's releases, the abilities of the most recent models continue to come on leaps and bounds. However, the fit of a tool to your application should extend beyond just which is the latest or has the best benchmarks. That said, all my Claude API calls come from Power Automate or Copilot Studio, demonstrating the ability of these platforms to bend to any tool choices you might wish to make.

Have you tried to GPT-5.1 yet, and if it's not your first choice yet, what models, platforms, or other tools are you spending most of your time with?

First posted on Linkedin on 11/13/2025 -> View Linkedin post here