When you open Copilot Chat in Microsoft 365, it uses a default model to generate responses. This default model may not always match your task. For example, the standard model might be too slow for quick drafts or lack the reasoning depth needed for complex data analysis. Changing the default model lets you pick a faster or more capable version based on your current work. This article explains how to switch the active model in Copilot Chat and what each model option does.
Key Takeaways: Switching Models in Copilot Chat
- Copilot Chat model selector: Located in the top-right corner of the chat pane, lets you switch between GPT-4 and GPT-4 Turbo instantly.
- GPT-4 (Default): Offers balanced speed and accuracy for general business tasks like drafting emails, summarizing documents, and answering questions.
- GPT-4 Turbo: Faster response time and lower latency, best for iterative conversations, code generation, and real-time collaboration.
What the Default Model in Copilot Chat Controls
The default model in Microsoft 365 Copilot Chat determines the underlying language model that processes your prompts. Two primary models are available: GPT-4 and GPT-4 Turbo. GPT-4 is the standard model optimized for accuracy and depth. It works well for tasks requiring thorough analysis, such as comparing financial data across multiple documents or generating detailed reports. GPT-4 Turbo is a faster variant with lower latency. It is designed for high-frequency interactions where speed matters more than absolute depth, such as real-time brainstorming, code debugging, or quick replies in a chat thread.
The model selection affects only the Copilot Chat experience within Microsoft 365 apps like Word, Excel, PowerPoint, and Teams. It does not change the model used by Copilot in other surfaces like the Edge sidebar or the standalone copilot.microsoft.com website. You must have an active Microsoft 365 Copilot license to access either model. Without a license, Copilot Chat uses a basic model with limited capabilities.
Steps to Change the Default Model in Copilot Chat
- Open Copilot Chat in a Microsoft 365 app
Launch any Microsoft 365 app such as Word, Excel, PowerPoint, or Teams. Click the Copilot icon in the ribbon or the floating chat button to open the Copilot Chat pane. - Locate the model selector
Look at the top-right corner of the Copilot Chat pane. You will see a dropdown menu labeled with the current model name, such as “GPT-4” or “GPT-4 Turbo.” - Click the dropdown to view available models
Click the model name. A short list appears showing the available models for your account. You will typically see “GPT-4 (Default)” and “GPT-4 Turbo.” - Select the desired model
Click the model you want to set as the default. The menu closes, and the model selector now shows your chosen model. All subsequent chat messages in this session use this model. - Confirm the change with a test prompt
Type a simple prompt such as “Summarize the last three sentences of this document” and press Enter. Observe the response speed and quality to verify the model switch took effect.
The model selection persists for the current session only. When you close and reopen the app, Copilot Chat resets to the tenant-level default model. To make the change permanent, your Microsoft 365 administrator must configure the default model in the Microsoft 365 admin center.
Common Issues When Changing the Default Model
Model selector is grayed out or missing
The model selector may appear grayed out or be completely absent. This typically means your Microsoft 365 tenant administrator has locked the model selection through a policy. Contact your admin and ask them to navigate to Microsoft 365 admin center > Settings > Org Settings > Copilot > Model selection. They can enable the “Allow users to choose their model” toggle. Once enabled, the selector becomes active.
GPT-4 Turbo option is not listed
If only GPT-4 appears, your tenant may not have access to GPT-4 Turbo. This model requires a specific licensing add-on or a Microsoft 365 Copilot plan that includes Turbo access. Check your subscription in the Microsoft 365 admin center under Billing > Licenses. If Turbo is not included, upgrade to a plan that supports it or wait for the admin to provision the add-on.
Model switch does not improve response speed
Switching to GPT-4 Turbo may not always yield faster responses if the network latency or server load is high. Try closing other tabs and apps to free bandwidth. If the issue persists, run a network speed test. A connection slower than 10 Mbps can negate the speed benefits of Turbo. For consistent performance, use a wired Ethernet connection or a 5 GHz Wi-Fi band.
Copilot Chat Model Options: GPT-4 vs GPT-4 Turbo
| Item | GPT-4 (Default) | GPT-4 Turbo |
|---|---|---|
| Response speed | Moderate, 2-5 seconds per prompt | Fast, 0.5-2 seconds per prompt |
| Context window | 8,192 tokens | 128,000 tokens |
| Best for | Deep analysis, document summarization, complex reasoning | Real-time chat, code generation, iterative drafting |
| Cost per token | Higher | Lower |
| Availability | All Microsoft 365 Copilot licenses | Requires specific Turbo add-on or higher-tier plan |
Choosing the right model depends on your task. For a quick email draft, GPT-4 Turbo saves time. For a multi-page contract review, GPT-4 provides more accurate clause analysis. Test both models on your typical workload to determine which one fits best.
You can now change the default model in Copilot Chat to match your task. Start by testing GPT-4 Turbo for quick replies and GPT-4 for detailed analysis. If the model selector is unavailable, ask your admin to enable user model selection. For advanced users, combine the model switch with the Copilot Chat plugin system to further customize responses. For example, use the SharePoint plugin with GPT-4 Turbo to retrieve and summarize documents in real time.