Understand consumption-based Copilot usage
Some Copilot experiences use a consumption model to measure the work they perform. These experiences track the work completed on your behalf. Usage-based charges can apply in addition to your license.
Understanding how usage accumulates helps you scope work effectively, choose the right approach for a task, and make informed decisions about how you use Copilot.
How consumption-based Copilot works
When you submit a request, Copilot performs work to achieve the outcome. Depending on the experience, that work may contribute to your organization's usage.
Usage can accumulate as Copilot:
- Retrieves and processes information
- Reads files and other data sources
- Reasons across multiple inputs
- Generates responses and drafted content
- Calls tools and connected capabilities
- Creates images or other media
- Performs actions in apps or browsers
The amount of usage depends on the work required to complete the task.
For example, a simple question that references a single source generally requires less work than a task that gathers information from multiple locations, analyzes findings, creates content, and completes follow-up actions.
The goal isn't to minimize usage. The goal is to use Copilot intentionally so that the value of the outcome matches the work required to produce it.
What affects how much a task uses
Several factors determine how much usage a task consumes. Knowing them helps you make intentional choices.
Task scope and complexity. The number of steps needed to achieve a goal is often one of the largest drivers of usage. For example, asking Copilot to summarize a document typically requires less work than asking it to analyze several documents, identify trends, and draft a recommendation report.
Information processed. The amount of information Copilot needs to analyze can significantly affect usage. Tasks that work with multiple files, messages, records, or data sources generally require more work than tasks focused on a single source.
Tools and capabilities. Many Copilot experiences can call tools, skills, or connected systems while completing a task. Searching the web, querying connected applications, generating images, or performing actions in other systems may increase the amount of work required to complete a request.
Model selection. Many consumption-based Copilot experiences let you choose the AI model powering a conversation. Model choice can affect response speed, depth, and credit usage. Total usage also depends on the context, tools, and work required.
Type of output generated. Some outputs require more work than others. For example, generating images or other media often requires more processing than generating text alone.
Work with usage in mind
Working efficiently doesn't mean asking Copilot for less. It means giving Copilot what it needs to produce a useful result without unnecessary rework.
- Be clear and specific. Clear requests reduce ambiguity and often lead to useful results with fewer revisions.
- Describe the desired outcome. Providing the overall goal helps Copilot understand what you're trying to accomplish and plan the work required to get there.
- Provide the right context. Including relevant files, details, or constraints can help Copilot produce better results without repeatedly gathering missing information.
- Use advanced capabilities intentionally. Features such as image generation, connected tools, and multi-step automation can deliver significant value when they're appropriate for the task.
- Review usage information when available. Some experiences provide visibility into usage. Reviewing this information can help you better understand how different tasks consume resources.
Remember: the objective is not to avoid usage. The objective is to achieve meaningful outcomes efficiently.
Apply these concepts in Copilot Cowork and Code
The principles you've learned apply across many consumption-based Copilot experiences. Copilot Cowork and Copilot Code both use Copilot Credits, which accumulate as these experiences perform work on your behalf.
In both experiences, higher-complexity work generally uses more credits than simpler requests.
Usage limits
Your organization sets how many credits you can use each month, and your limit might be part of a shared group limit. If you reach your limit, you can request more credits. The request goes to the person who manages licensing in your organization.
Check task usage
Type /cost in a conversation to view the approximate credits used by the current task. The command doesn't consume credits. It reports usage so far, not an advance price or an official billing record.
For example, Cowork might display:
610.6 credits used for this task so far.
Reviewing usage periodically can help you understand which types of tasks require relatively little work and which require more.
Review usage in one place
Depending on your organization's rollout, you might also see a personal usage view that combines Copilot Credit information across supported experiences such as Cowork and Code.
This view can include:
- Credits used to date
- Remaining credits
- Usage history over time
- The billing reset timeline
Use this view for trend monitoring and capacity planning. For task-level detail, use the usage indicators available in the experience you're working in, such as /cost in Cowork.
Choose a model
Cowork and Code can support multiple AI models.
For most work, leave the model selector set to Auto. Auto selects an appropriate model based on the nature of the task, helping balance capability and usage automatically.
More advanced models may provide additional reasoning capabilities for complex tasks, while simpler tasks might not require that level of capability.
To learn how the available models differ and when to change the selection, review the model guidance in the experience you're using.
Note
We recognize that different people like to learn in different ways. You can choose to complete this module in video-based format or you can read the content as text and images. The text contains greater detail than the videos, so in some cases you might want to refer to it as supplemental material to the video presentation.