Industry
Contents
Independent AI analysis, European lens

Microsoft Copilot Studio pricing and licensing explained

IC Guides · AI Agents · Work and business

Documentation researchedNo hands-on testMicrosoft Learn page dated 1 October 2026, checked 12 October 2026

Copilot Studio meters agent usage in Copilot Credits, and Microsoft publishes how many credits each feature uses. A classic answer is 1 credit, a generative answer 2, an agent action 5. Credits are pooled across the tenant, and when you reach 125% of prepaid capacity Microsoft disables custom agents. Staff who hold a Microsoft 365 Copilot licence use agents at no credit charge in employee facing scenarios, within fair use limits.

Who it fits

Organisations on Microsoft 365 who want agents in Teams, Outlook and SharePoint, or customer facing agents on the same platform. It is a build platform, so expect design and governance work. The credit price below comes from our tracker, because the licensing guide Microsoft links to is a separate document we did not read.

What Microsoft publishes

Agent featureCopilot CreditsAt $200 per 25,000 credits (our arithmetic)
Classic answer1$0.008
Generative answer2$0.016
Agent action5$0.04
Tenant graph grounding, per message10$0.08
Agent flow actions, per 100 actions13$0.104
Text and generative AI tools, basic, per 10 responses1$0.008
Text and generative AI tools, standard, per 10 responses15$0.12
Text and generative AI tools, premium, per 10 responses100$0.80
Content processing tools, per page8$0.064
Copilot Studio feature consumption in credits and Industry Contents’ conversion at the stated capacity price.
Voice tierCredits per minuteAt $200 per 25,000 credits
Classic voice10$0.08 a minute
GenAI voice35$0.28 a minute
Premium GenAI voice75$0.60 a minute
Voice-tier credit consumption per minute, converted at the capacity price stated in the guide.

Microsoft states the voice rates include core agent activity. The dollar columns are ours, from the tracker’s capacity price. They exclude bring your model configurations, which are billed separately.

Microsoft’s examples, in dollars (our conversion)

ExampleMicrosoft’s credit calculationDollars at $0.008 per credit
Website support agent with 4 classic and 2 generative answers per run, 900 customers a day[(4×1)+(2×2)] × 900 = 7,200 credits a day$57.60 a day, about $1,728 over 30 days
Sales agent with tenant graph grounding, using 4 generative answers plus 4 grounding, 100 unlicensed users[(4×2)+(4×10)] × 100 = 4,800 credits a day$38.40 a day
Order processing agent with 4 agent actions per order4 × 5 = 20 credits$0.16 per order
Microsoft’s examples, in dollars (our conversion). The table sets out Example, Microsoft’s credit calculation, Dollars at $0.008 per credit.

Microsoft notes a reasoning model adds the premium tool rate of 10 credits per 1,000 tokens on top of the feature rate, so reasoning can multiply the cost of a single answer.

What happens at the limit

At 125% of prepaid capacity, custom agents are disabled. A conversation in progress is not interrupted, but later attempts are rejected until capacity is raised or resets. Agent flows behave differently. When prepaid capacity is used up, new flow runs are blocked while the agent keeps working for other interactions, and enforcement resets monthly.

Pay as you go, linked to an Azure subscription, avoids enforcement because overage bills to Azure. Allocation also changes availability because an environment with its allocation keeps working when the tenant pool is exhausted, as Microsoft’s four environment example shows. Admins can set monthly limits per agent in the Power Platform admin center.

What is included for Microsoft 365 Copilot licence holders

Employee facing use is at no charge when the user holds a Microsoft 365 Copilot licence and the agent runs under that user’s identity, within fair usage limits Microsoft reserves the right to change. Microsoft lists two exceptions. Computer Using Agents are not included, and agent flows are free only when triggered by the “When an agent calls the flow” trigger. Other triggers use credits at the standard rate.

Where costs surprise you

One answer can draw several meters. A grounded generative answer used 12 credits in Microsoft’s example, six times a plain generative answer. Credits are also pooled, so one busy agent can exhaust capacity other teams planned on, unless you allocate by environment.

The agent can also switch off, so the 125% rule is a business continuity decision to make before launch and not after. A mixed audience of licensed and unlicensed users makes the bill depend on who asks.

How to test it

  1. Build the agent and run the Microsoft Copilot Studio agent usage estimator with your traffic assumptions.
  2. Run a pilot and read consumption by agent in the Power Platform admin center.
  3. Compare measured credits per conversation with the estimator, then multiply by your volume.
  4. Set per agent monthly limits and decide in advance what should happen at 125%.

Questions to ask before you buy

  1. What is the current price per credit for our agreement, for packs and for pay as you go?
  2. Which of our users hold Microsoft 365 Copilot licences, and which scenarios are covered?
  3. How should we allocate capacity across environments?
  4. Do unused credits expire, and when does the pool reset?
  5. What happens to customer facing agents at the 125% threshold, and can we opt for pay as you go instead?

Sources

Dollar figures are our arithmetic from the tracker’s capacity price. We did not test the product. Send a correction.

IC Lab · AI Agent Pricing Tracker

Vendor prices change. The tracker records when.

Every entry links to the vendor’s pricing page and shows the date we checked it. Open the tracker, or get major pricing changes by email every Friday in TalkHead.

Related guides Copilot Studio governance · Copilot Studio vs Agentforce · Salesforce Agentforce · ChatGPT workspace agents · AI agent glossary · AI Agent Pricing Tracker · All AI agent guides

Explore the next question

Why AI workers challenge traditional software pricing Read the analysis behind the changing unit economics. Compare the vendor pricing models Check the billing units, scenarios and primary sources before choosing.