Summary
Getting a Sui RPC endpoint in 2026 means choosing between a public fullnode URL, a managed RPC API, or a dedicated Sui node, then wiring the correct HTTPS endpoint into your SDK, wallet, or indexer. The right choice depends on whether you are prototyping, running production traffic, or need archive and high-throughput access.
This page walks through Sui endpoint formats, chain settings, a quick connectivity test, and the operational checks that separate a throwaway endpoint from one you can run in production. It also shows where OnFinality's Sui RPC API and dedicated node options fit into that decision.
Sui is a high-throughput Layer 1 where most application logic runs through an RPC endpoint: reading objects, submitting transactions, querying events, and streaming checkpoints. If you searched for how to get a Sui RPC endpoint in 2026, you probably need one of three things: a URL to paste into your SDK, a production-grade endpoint that will not buckle under load, or a way to test against Sui testnet before you ship. This page answers all three, then helps you decide which endpoint type actually fits your workload.
Which Sui endpoint type fits your workload?
Before you copy a URL, decide what you are connecting. The three common options behave very differently once traffic grows.
| Endpoint type | Best for | Watch out for |
|---|---|---|
| Public fullnode | Quick scripts, learning the API, one-off reads | Shared capacity, rate limits, no support path |
| Managed RPC API | Production apps, wallets, backends, indexers | Confirm archive depth, WebSocket, and rate policy |
| Dedicated Sui node | High-throughput or latency-sensitive workloads, strict isolation | Higher cost, needs an ops owner |
If you are prototyping, a public endpoint is fine. The moment you have real users, scheduled jobs, or a transaction pipeline, move to a managed RPC API. If your workload is steady and heavy enough that shared capacity becomes a bottleneck, a dedicated node gives you isolated resources and predictable behavior. OnFinality offers both a managed Sui RPC API and dedicated Sui nodes, so you can start managed and graduate to dedicated without changing your application code.
Sui endpoint formats and chain settings
Sui exposes an HTTPS JSON-RPC interface. The endpoint is a base URL; you append the method call in the request body, not the path. A typical Sui RPC URL looks like https://<host>/ and you POST JSON-RPC 2.0 payloads to it.
Key settings to keep straight:
- Transport: HTTPS for request/response. Sui also supports WebSocket subscriptions for event and checkpoint streams on providers that enable it.
- Network: Sui Mainnet and Sui Testnet are separate endpoints. Never point a production key at testnet.
- Chain identifier: Sui uses its own object and transaction model, so do not assume EVM-style
eth_*methods. Use Sui JSON-RPC methods such assui_getObject,sui_getLatestCheckpointSequenceNumber, andsui_executeTransactionBlock. - API versioning: Sui has been evolving its RPC surface. Pin your SDK version and check method availability against the endpoint you choose.
For the current OnFinality Sui endpoint and supported transports, see the Sui network page. That page is the authoritative source for the exact URL and any network-specific notes, so treat it as your reference rather than hardcoding a URL from a blog post.
Test the endpoint before you build on it
The fastest way to confirm a Sui endpoint works is a single JSON-RPC call. This request asks for the latest checkpoint sequence number, which any healthy fullnode should answer.
curl -s https://<your-sui-rpc-endpoint>/ \
-H 'Content-Type: application/json' \
-d '{
"jsonrpc": "2.0",
"id": 1,
"method": "sui_getLatestCheckpointSequenceNumber",
"params": []
}'
A healthy response returns a JSON object with a result field containing a numeric checkpoint. If you get a timeout, a 429, or an empty body, the endpoint is either rate-limited, misconfigured, or down. Replace <your-sui-rpc-endpoint> with the URL from the Sui network page.
Once the raw call works, wire it into your client. In a JavaScript/TypeScript app using the Sui SDK, you point the client at your endpoint:
import { SuiClient, getFullnodeUrl } from '@mysten/sui/client';
// Use your managed or dedicated endpoint in production.
const client = new SuiClient({
url: process.env.SUI_RPC_URL, // e.g. your OnFinality Sui endpoint
});
const checkpoint = await client.getLatestCheckpointSequenceNumber();
console.log('latest checkpoint:', checkpoint);
Keeping the URL in an environment variable lets you swap between public, managed, and dedicated endpoints without a code change.
Sui testnet: separate endpoint, separate keys
Sui Testnet is a distinct network with its own endpoint and its own faucet. If you are building or testing, request testnet SUI from the official Sui faucet, then point a test client at the Sui Testnet endpoint. Keep testnet and mainnet credentials completely separate: a key that can move mainnet assets should never be loaded into a test script.
A practical workflow:
- Develop against Sui Testnet using a managed testnet endpoint.
- Run integration tests that submit real transactions on testnet.
- Promote the same code to Sui Mainnet by changing only the endpoint URL and key.
OnFinality lists Sui Testnet alongside mainnet in its supported RPC networks, so you can keep both environments on the same provider and the same SDK configuration.
Public, managed, and dedicated: what changes at scale
A public fullnode is a shared resource. It is excellent for learning and light reads, but it is not designed for sustained throughput, and you have no support channel when it degrades. A managed RPC API adds capacity planning, monitoring, and a support path on top of the same JSON-RPC interface. A dedicated node gives you isolated compute and memory so that a heavy sui_getEvents query or a burst of transaction submissions does not compete with other tenants.
Signs it is time to move up a tier:
- You see intermittent 429 responses or latency spikes during peak hours.
- Your indexer or backend needs consistent throughput, not best-effort.
- You need archive-depth queries or WebSocket subscriptions that public endpoints do not guarantee.
- You want an SLA-style relationship and a named support contact.
If you are unsure which tier matches your traffic, the RPC pricing page outlines the options, and the provider selection guide covers the evaluation criteria in depth.
Common Sui RPC failure modes and how to debug them
Most Sui RPC problems fall into a handful of patterns. Match the symptom to the likely cause before you change providers.
| Symptom | Likely cause | First fix |
|---|---|---|
| 429 Too Many Requests | Shared endpoint rate limit | Reduce polling, batch reads, or move to managed/dedicated |
| Timeouts on large queries | Query too broad or endpoint under load | Paginate, narrow the object/event filter |
| Method not found | SDK/endpoint version mismatch | Check method availability and pin SDK version |
| Empty or stale data | Wrong network (testnet vs mainnet) | Verify the endpoint URL and network |
| WebSocket disconnects | Idle timeout or unstable connection | Add reconnect logic with backoff |
A useful habit is to log the endpoint host alongside every error. When you run multiple environments, most "Sui is broken" reports turn out to be a testnet URL in a mainnet config.
Production readiness checklist
Before you point real traffic at a Sui endpoint, confirm each of these:
- Endpoint ownership: You know which provider and tier you are on, and you have a support path.
- Environment separation: Mainnet and testnet URLs and keys are stored separately.
- Failover: You have a second endpoint or provider to switch to if the primary degrades.
- Monitoring: You track error rate, latency, and rate-limit responses, not just uptime.
- Archive needs: If you query historical state, confirm the endpoint supports the depth you need.
- Subscriptions: If you rely on event streams, confirm WebSocket support and test reconnection.
- Cost model: You understand how request volume maps to your bill.
If any of these are unclear, that is the signal to move from a public endpoint to a managed or dedicated one. OnFinality's RPC API and dedicated node options are built for exactly this transition.
Key Takeaways
- A Sui RPC endpoint is an HTTPS JSON-RPC base URL; you POST JSON-RPC 2.0 payloads to it, not EVM-style methods.
- Sui Mainnet and Sui Testnet are separate endpoints with separate keys and faucets.
- Public fullnodes are fine for prototyping; managed RPC APIs and dedicated nodes are for production traffic.
- Test any endpoint with a single
sui_getLatestCheckpointSequenceNumbercall before building on it. - Most Sui RPC failures are rate limits, version mismatches, or the wrong network, not a broken chain.
- Keep the endpoint in an environment variable so you can switch tiers without code changes.
- See the Sui network page for the current endpoint and RPC pricing for tier options.
Frequently Asked Questions
Is there a free Sui RPC endpoint? Yes. Public fullnodes are available for light use and testing. They are shared, so they are not ideal for production throughput. For sustained traffic, use a managed RPC API or a dedicated node.
What is the difference between Sui Mainnet and Sui Testnet endpoints? They are separate networks with separate state and separate endpoints. Testnet SUI comes from a faucet and has no real value. Always keep the two configurations apart.
Does Sui use JSON-RPC or gRPC? Sui exposes a JSON-RPC interface over HTTPS, and some providers also offer WebSocket subscriptions. Check the specific endpoint's supported transports before you design around one.
How do I know if my Sui endpoint is rate-limited? You will typically see HTTP 429 responses or sudden latency spikes under load. If that happens during normal traffic, it is time to move to a managed or dedicated tier.
Can I use one endpoint for both mainnet and testnet? No. They are different networks. Use separate endpoints and separate credentials for each environment.
Where do I find the current OnFinality Sui endpoint? The Sui network page lists the endpoint and supported transports. Use that as your source of truth rather than a hardcoded URL from any article.
Next steps
If you are still prototyping, grab a public endpoint and run the checkpoint test above. If you are heading to production, compare managed and dedicated options on the RPC pricing page, then review the full list of supported RPC networks to confirm Sui and any other chains you depend on are covered. When you are ready, OnFinality's Sui RPC API gives you a managed endpoint you can point your SDK at today, with a path to dedicated Sui nodes as your traffic grows.