> ## Documentation Index > Fetch the complete documentation index at: https://docs.openlayer.com/llms.txt > Use this file to discover all available pages before exploring further. # Plans and usage > Learn what your workspace plan can limit, how requests and other usage are counted, and where to see your usage. Your plan applies to the whole workspace. It sets limits on resources, tracks usage, and turns modules and features on or off. ## What a plan can limit ### Resources | Limit | What it counts | | - | - | | Members | People in the workspace. | | Projects | Projects in the workspace. | | Data sources / project | [Data sources](https://docs.openlayer.com/glossary.md#data-source) in one project. | | Development tests / project | Tests in one project's [Development](https://docs.openlayer.com/development/overview.md) mode. | | Monitoring tests / project | Tests that run on production data in one project. | | Custom metrics / project | [Custom metrics](https://docs.openlayer.com/tests/custom-metrics.md) in one project. | | Rows per dataset | Rows in each dataset of a [Development](https://docs.openlayer.com/development/overview.md) commit, training and validation alike. | | Max data retention (days) | The longest [retention period](https://docs.openlayer.com/workspace-and-projects/data-retention.md) you can set. | | Active frameworks | Activated [frameworks](https://docs.openlayer.com/glossary.md#framework) in the workspace, built-in ones included. | | Custom frameworks | [Frameworks the workspace created](https://docs.openlayer.com/governance/build-custom-framework.md). | | Custom rules | [Rules](https://docs.openlayer.com/glossary.md#rule) the workspace created. Rules that come with a built-in framework don't count. | A Development commit with a dataset over the rows per dataset limit fails during processing. The error names the dataset, its row count, and the limit. Activating a framework over the active frameworks limit fails. Creating a custom framework or custom rules over their limits fails too, and a batch of rules counts every rule in it. ### Usage A [request](https://docs.openlayer.com/glossary.md#request) is one trace sent to Openlayer. Each trace counts as one request. | Limit | Behavior | | - | - | | Requests / month | Counted per calendar month. On self-serve plans, such as the free plan, it blocks new requests once reached. | | Commit processing time / month | Time spent processing [Development](https://docs.openlayer.com/development/overview.md) commits. Once used up, new commits are refused until next month. | | Custom metric processing time / month | Time spent running [custom metrics](https://docs.openlayer.com/tests/custom-metrics.md). Once used up, a commit's custom metric run fails with the reason, and Monitoring and Development evaluations skip the tests that need custom metrics until next month. | | Requests / year | Tracked and notified. Never blocks. | | Tokens / year | Tracked and notified. Never blocks. | | Scorer runs / year | Tracked and notified. Never blocks. | | Test executions / year | Tracked and notified. Never blocks. | | Storage | Tracked and notified. Never blocks. | A [scorer run](https://docs.openlayer.com/glossary.md#scorer-run) is one LLM-judge test evaluated on one request. A test execution is one test evaluated once, passing or failing, on a commit or a Monitoring [evaluation window](https://docs.openlayer.com/monitoring/evaluation-and-delay-windows.md#evaluation-windows). Skipped and errored tests don't count. A scorer run counts each request scored, while a test execution counts each test result. Yearly usage covers the last 12 calendar months, including the current month. A plan with a monthly request limit also has a yearly one of 12 times it, unless the plan sets its own. A yearly limit doesn't imply a monthly one. Monthly limits, including both processing times, reset each calendar month. Storage is the size of the requests received, measured as JSON, plus commit artifacts. It counts requests from the start of the longest [retention period](https://docs.openlayer.com/workspace-and-projects/data-retention.md) among the workspace's projects. If a project keeps data indefinitely, it counts every request. Shortening retention lowers storage. ### Streaming rate A plan can set how many [data stream](https://docs.openlayer.com/api-reference/rest/monitoring/stream-data.md) requests per second each data source accepts. Without one, the deployment default applies: 10 per second, unless your deployment configures another rate. Requests above the rate get HTTP `429 Too Many Requests`. Retry them with backoff. The Python and TypeScript SDKs retry a `429` automatically, twice by default, with exponential backoff. ### Modules A plan can turn each module on or off: * [Governance + Compliance](https://docs.openlayer.com/governance/overview.md) * Monitoring, documented under [Observability](https://docs.openlayer.com/monitoring/overview.md) * Development, documented under [Offline testing](https://docs.openlayer.com/development/overview.md) * [Gateway](https://docs.openlayer.com/gateway/overview.md) * [Data quality monitoring](https://docs.openlayer.com/data-quality-monitoring/overview.md) * [Traditional ML](https://docs.openlayer.com/monitoring/publishing-tabular-predictions.md) Data quality monitoring and Traditional ML gate non-LLM projects: * Creating a classification project needs Traditional ML. * Creating a tabular regression project needs either Traditional ML or Data quality monitoring. * Adding a streamed data source to a non-LLM project needs Traditional ML. * Adding a warehouse-backed data source (BigQuery, Snowflake, Databricks, or Postgres) to a non-LLM project needs Data quality monitoring. ### Features A plan can also turn these features on or off: * AI assistant * [SAML SSO](https://docs.openlayer.com/security/saml-sso.md) * [Data export](https://docs.openlayer.com/monitoring/export-data.md) * Data annotation * [Role-based access control](https://docs.openlayer.com/security/roles-and-permissions.md) ## Trial New workspaces start with a 14-day trial of the Pro plan, with every module and feature. When the trial ends, the workspace is locked in the app until you upgrade or extend the trial. Your projects and data are kept. To upgrade or extend, [contact sales](https://www.openlayer.com/contact). ## Free plan Workspaces on the free plan include Monitoring and Development, and keep data for up to 90 days. The other modules, the AI assistant, custom metrics, data export, data annotation, and role-based access control are on paid plans. Workspaces created on the free plan before these limits keep the features they already had. ## See your usage and limits * **Settings → Billing** shows your plan for the whole workspace. Usage cards show where you stand against each limit, next to your yearly usage, test executions included, and the modules your plan includes. The **Limits** section lists the per-project limits, rows per dataset, data retention (**Up to N days** or **Any period**), and the streaming rate per data source. The **Usage** section shows commit and custom metric processing time in hours, and storage. When your plan includes Governance + Compliance, it also shows cards for active frameworks, custom frameworks, and custom rules. * **Project settings → General** shows usage for one project, such as its data sources and tests. ## When you reach a limit When an action would go over a limit, Openlayer returns an error that names the limit you reached. For example, creating a data source in a project that already has the maximum fails with an error about data sources per project. Yearly usage limits and storage never block. Openlayer tracks them and notifies you instead. To change your plan or raise a limit, [contact sales](https://www.openlayer.com/contact). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.