Limits keep one workload from slowing down another. They are separate from
your plan’s monthly allowances (see
Plans and limits) and from the
spend cap.
Data API, Bolt and MCP
Per organization, across all its databases, keys and protocols:
| Plan | Requests per second | Burst | Concurrent queries | Query timeout | Open Bolt sessions |
|---|
| Free | 25 | 50 | 8 | 10 s | 64 |
| Pro | 250 | 500 | 64 | 30 s | 512 |
| Scale | 1,000 | 2,000 | 256 | 60 s | 2,048 |
| Enterprise | set in your agreement | | | | |
- Requests per second is a token bucket: you can burst up to the burst size,
then sustain the rate. Each API key also has its own bucket of the same
size.
- Over a limit:
429 rate_limited with Retry-After (over Bolt, a transient
error that drivers retry). Requests refused this way are not counted as
usage.
- Client addresses that present many invalid keys are rate limited too.
| Size limit | Value |
|---|
| HTTP request body | 2 MiB (413 payload_too_large) |
| Result | 64 MiB (413 result_too_large) |
| Bolt connections per client address | 512 |
| Bolt handshake and authentication | 10 s |
| MCP JSON-RPC batch | 32 messages |
Staying under the limits
- Reuse connections: one SDK client or driver instance per process.
- Batch writes with
UNWIND $rows (the SDKs’ bulk helpers do it for you)
instead of one request per row.
- Back off on
429 as Retry-After says; the SDKs do this automatically.
- Long analytical queries hold a concurrency slot for their whole run; use a
clone for heavy analytics.
Management API
| Limit | Value |
|---|
| Requests per token | 300 per minute (X-RateLimit-Limit, X-RateLimit-Remaining) |
| Request body | 64 KiB |
| Database deletions | 20 per organization per hour |
| On-demand, clone and point-in-time backups | 30 per organization per hour |
| Restores, clones and point-in-time restores together | 20 per organization per hour |
| Active management tokens | 25 per organization |
| Invitations | 30 per organization per hour |
| Exports | 20 per organization per 24 hours; one per database at a time |
These budgets apply to the console too.
Other limits
| Limit |
|---|
| Active API keys | 50 per organization |
| Active management tokens | 25 per organization |
| SCIM tokens | 5 active per organization |
| Webhooks | 10 per organization |
| Saved queries | 200 per organization |
| IP allowlist entries | 1 / 20 / 100 per list (Free / Pro / Scale) |
| Audit log CSV export | 100,000 rows per file, 30 exports per hour |
| Console import file | 10 MB / 100 MB / 250 MB (Free / Pro / Scale) |