Reference

Plans and limits

Limits stop growth, never existing use. Reading published configuration and evaluating flags are never limited and never billed per evaluation.

Plans

Plan	Limits
Explore	1 project 3 members 3-day diagnostics retention 10,000 sampled events per day per environment
Team	3 projects 15 members 14-day diagnostics retention 50,000 sampled events per day per environment
Scale	10 projects Unlimited members 30-day diagnostics retention 500,000 sampled events per day per environment
Enterprise	Custom projects, members and retention

Prices and the full comparison are on the pricing page. An organization without an active, trialing or past-due subscription is on Explore.

Features by plan

Feature	Explore	Team	Scale
Required independent approval	Not included	Included	Included
Required scenario checks	Not included	Included	Included
CI run results	Not included	Included	Included
Webhooks	Not included	Included	Included
Scheduled publishing	Not included	Included	Included
Shared test sessions	Not included	Not included	Included
Audit log CSV export	Not included	Not included	Included

Every plan includes flags, segments, environments and revisions, changes, scenarios and diagnostics, and local evaluation in the SDKs.

When a limit is reached

Projects and members: creating another project, adding a member or sending an invitation is refused with LIMIT_REACHED (pending invitations count as seats). Existing projects and members are unaffected.
Plan features: on a plan without a feature, the dashboard does not let you turn it on or create new items (PLAN_REQUIRED). Existing data stays readable; for example, webhooks added earlier keep delivering until you disable or delete them, and test sessions created earlier keep their expiry and can still be revoked.
Diagnostics retention: sampled events are kept for the plan's retention (Explore 3 days, Team 14 days, Scale 30 days) and deleted by a job that runs every hour. See Data retention.

Reads are never blocked

No plan limit affects /v1/bundle, /v1/live, /v1/stream, /v1/manifest, or evaluation in your processes. Evaluation is local and synchronous, so the number of evaluations is neither limited nor billed. Reaching the daily event cap only stops storing sampled events; flags keep working exactly as before.

Daily event cap

Each environment has a daily cap on stored sampled events:

New environments start with a cap of 50,000 events per day. Admins change it per environment under Settings → Environments, up to the plan's figure (Explore 10,000 · Team 50,000 · Scale 500,000 per environment per day); 0 stores no events.
Once the day's stored events reach the cap, POST /v1/events still answers 202, with "capped": true and accepted: 0. Further sampled events are not stored until the next day (UTC). The SDK does not retry them.
Service reports are still recorded while capped, so "which revision is this service running" keeps updating.
The billing settings show each environment's usage against its cap.

While an environment is capped, diagnostics under-count evaluations for the rest of the day. Missing events are not evidence that a flag is unused.

Request limits

Request	Limit
`POST /v1/events`	64 KB body, 50 events
`PUT /v1/ci/code-refs`	4 MB body, 5,000 references, snippets cut to 240 characters
`POST /v1/ci/scenarios/run`, `POST /v1/admin/publish`	16 KB body
`GET /v1/live`	10,000 live sockets per environment (then 503 and the SDK falls back to the event stream and polling); revisions above about 1.9 MB are delivered by polling instead of the socket