Rate limits & tiers

Three things are metered: stream egress (GB delivered), RPC and /info request rate (req/s), and historical export volume. Limits are per key and set by your tier. Full pricing and the usage calculator live on /pricing.

Tier gates

Base price

Pro$49/mo
Ultra$199/mo

API keys

Pro5
Ultra25

Concurrent streams

Pro25
UltraUnlimited

Stream egress included

Pro50 GB/day
Ultra500 GB/day

Coin / wallet filters

Pro25 coins · 25 wallets
UltraUnlimited

Billing

ProMonthly or annual
UltraMonthly or annual

HyperEVM JSON-RPC

Pro250 req/s
Ultra1,000 req/s

eth_sendRawTransaction

Pro
Ultra

eth_getLogs

Pro
Ultra

HyperCore /info API

Pro250 req/s
Ultra1,000 req/s

Stream availability per tier (which streams, filter breadth, replay) and historical quotas are on the full pricing matrix.

What happens at the limit

The honest version: paid tiers are not hard-capped. Streams keep flowing and RPC keeps answering past your included quota; the overage bills at the published per-GB and per-million-request rates. No mid-session cutoff. Hard 429s only appear when you push past your plan's req/s rate; back off and retry.

SurfaceSignalMeaning
HyperEVM RPC / HyperCore /info / historicalHTTP 429Requests above your tier's req/s rate. Back off and retry; sustained 429s mean you need a higher rate, not faster retries.
gRPC streams (connect)RESOURCE_EXHAUSTEDSubscription rejected: concurrent-stream cap reached, or a filter (coins/levels) outside your tier's breadth.
gRPC streams (past plan rate, in-flight)throttled deliveryMessages are paced down rather than the stream being killed. Upgrade for full-rate delivery.
gRPC streams (falling behind)DATA_LOSSNot a rate limit: your consumer is slower than the feed. See the Operations guide for recovery.

Staying under your limits

  • Filter server-side. coins and wallets filters mean you only receive what you subscribe to, and only pay egress for that.
  • Take the narrowest stream that answers your question: choosing a stream maps needs to surfaces. A BTC ticker over the book-diffs firehose is the classic mistake.
  • Watch consumption in the console. Usage is metered per key, so one key per consumer tells you exactly where the gigabytes go.
  • If your estimate keeps crossing tiers, run the usage calculator; past a point, a dedicated node with private throughput is the better shape (Enterprise).