Power Platform Community Forum Thread Details

Hello,

Since I believe June 2026 the Copilot Studio portal offered a new experience with a new way of building agents that does not make use of conversation nodes (e.g. messages) / topics / prompts as everything is managed by the underlying generative orchestrator.

As such, I was wondering how the classic billing rate model is translated into the new experience, given the absence of said features and the introduction of new (skills, memory, ...).

Attached, the current billing model.

https://learn.microsoft.com/en-us/microsoft-copilot-studio/requirements-messages-management

Just a few considerations:

Classic answer and AI tools for example don't make sense to me as they represent message nodes and AI prompts which are not part of the new experience.
Is generative answer the new baseline message rate? Or will it be token based like prompts?

Thank you for your answers!

Categories:

General topics

Bot analytics

Bot administration

@DM-02040851-0

Yes, in the new experience, the concept of answer and AI tools tier to conversation noes and prompts have shifted. The reason is the underlying generative orchestrator treats the flow differently. Let's dive into-

In new experience, generative answers replaces the message nodes or answers. They represent the baseline generative AI reponses the agent produces. So, to answer your question directlym yes, generative answers are the new baseline unit for billing, kind of similar to the old message or prompt counts.

Is toke-based or per-message? The new experience meters primarly takene care of by tokens processed (I/P + O/P) withing generative answers. Kind of similar with how Anthropic and OpenAI models are billed.

AI tools and Skills - layered capabilities layered on to of generative answers. Hm.. additional expense for you/us but more precise and clean. Based on the AI and Tool ans Skills complexity the consumes credits, yes, distinct from the baseline generative cost. Let's consider an exmaple, calling an AI tool or invoking a skill may (must) add extra credits in addition of the generative answers tokens.

Not that important, but alwasy good to know, calls to memory or orchestration logic also consume credits, but these are separate from the generative answer cost.

Quick summary:

The new billing model is more granular and toke centric, reflects actual computational cost and AI usage (not just counting message). Generative answer billed by token processed, AI tools and skills will cut your pocket with additional cost.

References:

Managing AI experiences enabled by usage-based billing
Copilot Messages vs Copilot Credits — What Changed?
The four buying mechanims

I am sure some clues I tried to give. If these clues help to resolve the issue brought you by here, please don't forget to check the box Does this answer your question? At the same time, I am pretty sure you have liked the response!