⚙️ Your agent
Harness
Standard harness: rule-based agents and agent flows — billed by feature events (classic/generative answers, actions) after publish.
Audience & volume
Authenticated business-to-employee usage by M365 Copilot licensed users is zero-rated — no credits consumed (fair-use limits apply).
Per-conversation feature mix
Pre-authored responses — 1 credit each
AI-generated answers over knowledge sources — 2 credits each
Connectors, topic transitions, deep reasoning, computer use — 5 credits each
Adds 10 credits per grounded response (applied to generative answers)
Monthly extras
Billed at 13 credits per 100 actions. Flows triggered by "When an agent calls the flow" are zero-rated for M365 Copilot licensed users.
Document / image processing tools — 8 credits per page or image
📊 Monthly estimate
0 credits/mo
After contingency and M365 Copilot zero-rating
📏 Official Copilot Credits billing rates
Standard harness feature rates below. Agents on the GitHub Copilot harness are billed differently — consumption-based credits for LLM tokens, tools and the harness runtime, starting from build time (published planning ranges: ≈70–200 credits per light task, 400–600 medium, 1,500+ heavy). The Copilot chat harness (extending M365 Copilot Chat) is consumption-based or included in M365 Copilot licences.
| Agent feature | Credits |
|---|---|
| Classic answer | 1 |
| Generative answer | 2 |
| Agent action (incl. computer use, standard model) | 5 |
| Tenant graph grounding (per grounded response) | 10 |
| Agent flow actions (per 100) | 13 |
| Content processing tools (per page / image) | 8 |
| Text & generative AI tools — basic / standard / premium (per 1K tokens) | 0.1 / 1.5 / 10 |