Capacity
01- Shared RPC
- Shared pool, scaled by us
- Dedicated nodes
- Reserved node, yours alone
For developers
Developer API
for Web3 and blockchain apps
RPC, WebSocket subscriptions and a REST blockchain API from one place, under one key. Build the integration on the free plan, then move production traffic to the endpoint it needs — without running nodes in-house.
- 4
- Networks in production
- 58 ms
- Response, p95 across methods
- 100K
- Free requests a month
Project key
sk_live_9f2c7a41••••••••••••
- httpsrpc.settla.net/eth/mainneteth_blockNumber
- wssrpc.settla.net/eth/mainneteth_subscribe · newHeads
- httpsapi.settla.net/v1/market/quotesGET · symbols=BTC,ETH
One key for RPC, subscriptions and REST
Developer workflow
From API key to production endpoint
Four steps, and every one of them is a line of code rather than a ticket: the key works right after signup, the endpoint answers the first call, and the plan changes without touching the integration.
Your codeSettla APIProduction
01
Create a key
One key opens every network and every product. It travels in a header, not in the URL, so it never lands in logs and referrers.
x-api-key: sk_live_9f2c7a41
02
Send the first call
The same JSON-RPC the node itself speaks. No wrapper to learn, no schema of ours to map onto the chain.
POST /eth/mainnet
03
Subscribe to events
Blocks, logs and transfers arrive over one socket — instead of a worker asking the node the same question all day.
eth_subscribe · newHeads
04
Scale the endpoint
A plan raises the ceiling, a dedicated node removes the neighbours. The address and the key stay exactly as they were.
200 rps
Subscriptions
Subscribe instead of polling
One minute of the same network seen two ways: a worker that asks every few seconds, and a subscription that gets told. Pick a network — what polling costs is different in each one.
block every 12 s
Polling every 5 s
12 requests a minute
- 5 / 5
- Blocks seen this minute
- ≈ 2.5 s
- Time to find out
Subscription
one connection
- 5 / 5
- Blocks seen this minute
- 6 ms
- Time to find out
Here the worker does catch every block, but finds out about it 2.5 s late on average, and pays for the wait in requests.
A worker like that runs round the clock: 518,400 requests a month, or 5.2× the free plan's monthly limit. A subscription holds one connection and hears about the same blocks first.
Shared or dedicated
Start shared. Move to dedicated when the queue starts costing you.
Both run on the same API and the same key, so moving over is a URL change rather than a rewrite.
| Criterion | Shared RPC | Dedicated nodes |
|---|---|---|
| Capacity | Shared pool, scaled by us | Reserved node, yours alone |
| Noisy neighbours | Fair queue per key | None — single tenant |
| Rate limits | Per plan, burst-friendly | Whatever the node can serve |
| Archive | Included | Included, custom retention |
| Placement | Nearest healthy region | Pinned to your region |
| Setup | Instant | Provisioned within a day |
| Best for | Wallets, apps, bursty traffic | Exchanges, indexers, steady load |
Noisy neighbours
02- Shared RPC
- Fair queue per key
- Dedicated nodes
- None — single tenant
Rate limits
03- Shared RPC
- Per plan, burst-friendly
- Dedicated nodes
- Whatever the node can serve
Archive
04- Shared RPC
- Included
- Dedicated nodes
- Included, custom retention
Placement
05- Shared RPC
- Nearest healthy region
- Dedicated nodes
- Pinned to your region
Setup
06- Shared RPC
- Instant
- Dedicated nodes
- Provisioned within a day
Best for
07- Shared RPC
- Wallets, apps, bursty traffic
- Dedicated nodes
- Exchanges, indexers, steady load
Questions
What developers ask first
JSON-RPC over HTTPS, subscriptions over WebSocket, and a REST API for balances, transfers, fees and market data. Which methods a chain exposes depends on the chain itself — they are listed per network on the shared endpoints page.
That is what the Start plan is for: it is free, it comes with a key, and the limit is enough to build the integration and run it against a few methods. Production traffic moves to the right endpoint type afterwards, without a second integration.
Yes, wherever the network itself supports them: new blocks, logs and slots. The address is the same one you use for RPC — only the scheme changes, wss:// instead of https://.
The limit is counted per key: requests a month and a ceiling per second. Usage and overage are visible in the dashboard — by day, by key and by node. A ceiling that has become tight is a reason to change the plan, not the code — the endpoint and the key stay the same.
Yes, and they are shaped the same way: same host, same key, only the network segment of the path changes. Moving from a testnet to mainnet is one line of configuration, not a second integration.
When what you run into is not the plan but the neighbours: you need a flat latency tail, heavy archive queries or methods the shared endpoints do not carry. The table above compares both side by side.
Infrastructure shouldn't be the thing slowing you down.
Free to start. No credit card required.