Limits & Fair Use
All HighTower endpoints enforce rate limits at the IP level. The specific caps depend on your subscription tier and the interface you're using.
HTTP JSON-RPC
| Tier | RPS (read requests) | TPS (sendTransaction) |
|---|---|---|
| Free | 20 | 10 |
| Build | 100 | 50 |
| Grow | 150 | 100 |
| Accelerate | 400 | 100 |
WebSocket
| Metric | Shared Tiers | Dedicated |
|---|---|---|
| Concurrent connections | 10 | Custom |
| Subscriptions per connection | 50 | Custom |
| Message rate | Fair use | Uncapped* |
Yellowstone gRPC
| Tier | Concurrent Streams | Message Rate |
|---|---|---|
| Free | — (no access) | — |
| Build | 2 | Fair use |
| Grow | 5 | Fair use |
| Accelerate | 10 | High priority |
| Dedicated | Unlimited* | Uncapped* |
*Hardware limits apply.
Individual gRPC messages are capped at 4 MB. Use account/program filters to keep responses within this size.
What Happens When You Hit a Limit
- HTTP/gRPC: the server returns 429 Too Many Requests. Back off exponentially before retrying.
- WebSocket: the oldest connection or subscription is dropped. Your client should detect the disconnect and reconnect with backoff.
- Sustained abuse: if traffic patterns consistently overload shared infrastructure, our team will reach out via Discord to discuss options (tier upgrade or dedicated node).
Fair Use
There are no hidden quotas beyond the numbers listed above. "Fair use" on gRPC message rate means we don't enforce a hard cap per second, but we monitor aggregate bandwidth. If a single client's traffic visibly degrades service for others on the same shared node, we'll throttle that client and open a conversation about moving to dedicated hardware.