# Budgets and metering Source: https://docs.coderhelm.com/plugins/budgets Every model call a plugin makes is metered and counted against its own monthly limit and the team's. > Coming soon: this page describes work that is not on main yet. Plugins are being built and switched on team by team; they are not on main yet. This page follows the work in progress and will be updated when plugins ship. Every model call a plugin run makes is metered, like every other call CoderHelm makes, and paid for with the team's own model key. ## A limit for the plugin [#a-limit-for-the-plugin] ```yaml budget: monthly_tokens: 2000000 ``` `monthly_tokens` is the plugin's own monthly limit. The team can change it on the plugin's page under **Monthly token limit**, where it also sees **Tokens used this month**; empty or 0 means no limit of its own. The plugin's usage counts toward the team's [token limits](/concepts/token-limits) too. When either is reached, the plugin stops starting runs until next month, and a run it would have started shows as refused with the reason. ## Keep runs cheap [#keep-runs-cheap] - Put stable text (brand guide, checklists) in skills: on later calls it is read from the prompt cache, which costs far less than new input. - Tell scheduled agents to stop early when there is nothing to do. - Allow only the tools the plugin needs; every tool's description is part of the prompt.