Credits and usage

What a credit is

One credit is one cent of model cost. Every call to a model is priced from its real token cost: input, output, and cached input at a lower rate. A short chat message costs a fraction of a credit; a long document or a multi-step automation run costs more. The Insights tab of an automation shows what its runs cost.

One pool

AI chat, automations and media generation share one monthly pool per user in a workspace. The counter resets on the first of each month. The meter at the bottom of the sidebar shows used / limit; it turns amber at 80 percent and red at the limit. Account → Usage this month shows the same with the renewal date.

Plans

Your plan sets the monthly credits per user; the meter shows your limit. Every plan includes the full product: unlimited automations and workflows, unlimited meeting recordings, all integrations, shared workspaces with roles, and every frontier model. Only usage differs. Plans and prices are on the pricing page.

What costs nothing

• Two small models are free and cost no credits.
• Reading knowledge, running compiled code steps, and calling connected apps cost no credits by themselves; only the model calls around them do.

When credits run out

Chat and automation runs fall back to the free model, so you are never locked out and no run is lost. Steps that need a stronger model may give weaker results until the counter resets or the limit is raised. Nothing is deleted.

Keeping costs down

• Let Auto route. It sends easy messages to cheaper models.
• Compile automations fully. A plan where every step has a kind runs as a program and calls a model only where judgement is needed.
• Keep workspace Instructions short. They are sent with every message.