How do agent credits and usage work?

Production agents are usage-billed; free workspaces get test-only agent credits.

Free workspaces get test-only agent credits so you can try Loop Agent and MCP. Production agent usage is billed on paid plans, separately from your seat price.

See usage for this month

1. Open Settings → Reporting → Agent usage (owners and admins).
2. Review Runs, Replies, and the Daily runs chart.
3. Scroll to By agent and Recent activity to see which agents ran, which model they used, and token counts.

On iOS this screen is read-only. Spend limits and payment methods are edited on the web.

Free vs production

1. On Free, usage shows activity only — not dollars. Credits are for trying agents, not running them in production.
2. When included activity is used up, FabricLoop shows: "You have used the agent activity included on the free plan. Upgrade to keep your agents running and add more of them."
3. Tap View upgrade options (or open pricing) to move to Standard or Enterprise.

Paid usage and spend limits

1. On Standard or Enterprise, agent model costs (Anthropic Haiku / Sonnet / Opus, or ChatGPT Sol / Terra / Luna) are billed as you go — not included in the $10 / $20 seat price.
2. Open billing on the web to see month-to-date spend versus an optional monthly limit.
3. Admins can set a monthly spend limit and choose to hard-stop new agent work when the limit is reached. Raise the limit there to resume.

Tips

• You never pick a seat count for agents — they are not billed as members. See Who counts as an active member?.
• iOS links to Manage billing on the web from Agent usage when you are on a paid plan.
• For a workspace-wide cap, see Can I limit AI usage for my workspace?.

Related

• Can I limit AI usage for my workspace?
• What is included in the free plan?
• How does FabricLoop billing work?
• AI & Agents

Still stuck?

Contact us or browse all help topics.