Sui RPC, October 29, 2026
Changelog: October 29, 2026 (r3). Editorial pick puts Chainstack second for GraphQL and flat RU pricing; latency sample order unchanged (QuickNode still second on ms). Earlier: dual lists, Alchemy CU docs, anchors, BlockEden, column balance.
The best Sui RPC provider in 2026 is Tatum for production apps that need gRPC and multi-region latency. Editorial top five: Tatum, Chainstack, QuickNode, Ankr, Alchemy. Pure latency order in the five-city sample: Tatum, QuickNode, Alchemy, Chainstack, Ankr. Those lists diverge on purpose: Chainstack is editorial #2 for GraphQL and flat RU pricing even though it is fourth on latency.
Sui has retired JSON-RPC on its own public full nodes (week of July 27, 2026; code removal planned mid-October 2026). gRPC is the interface that remains. Latency figures below come from Tatum's published RPC benchmark (August 28, 2025; rechecked October 29, 2026). Treat them as one public table, not a third-party audit. Directory listings on CompareNodes are context, not a score.
Best for (quick picks)
- Best overall (editorial): Tatum. gRPC, JSON-RPC gateway, 18 to 33 ms in the sample.
- Latency order (sample only): Tatum (18 to 33 ms), QuickNode (39 to 70), Alchemy (55 to 96), Chainstack (72 to 112), Ankr (129 to 205).
- Best latency among peers: QuickNode (39 to 70 ms; next after Tatum in every city).
- Best GraphQL + flat RU pricing: Chainstack (mainnet GraphQL; free 3M RU/month).
- Best large free credit pool: Ankr (200M API credits/month on Freemium; GraphQL is Premium).
- Best if you already run Alchemy: Alchemy (mainnet gRPC; no GraphQL on Sui's table).
- Also consider for three networks: BlockEden.xyz (devnet, mainnet, and testnet on CompareNodes; JSON-RPC focused).
Best for: production apps that need gRPC, a JSON-RPC gateway during migration, and multi-region latency.
gRPC on mainnet, testnet, and devnet; seven documented services; same account for indexed data and webhooks. Lowest latency in the five-city sample (18 to 33 ms). Free plan: 100K lifetime credits at 3 RPS. Paid Starter from $49/month with 4M credits and 200 RPS (checked October 29, 2026).
Trade-off: GraphQL is not in the Tatum Sui reference. Tatum is not on Sui's own RPC provider table (checked October 29, 2026).
Best for: GraphQL on mainnet with flat 1 RU per request, including GraphQL.
On Sui's provider table for gRPC (testnet and mainnet) and GraphQL (mainnet); archival marked N/A there. Free Developer plan: 3M RU/month at 25 RPS. Growth: $49/month for 20M RU at 250 RPS. Extended history on Chainstack's Sui docs (not a full genesis archive).
Trade-off: fourth on latency in every city in the sample (72 to 112 ms). GraphQL is mainnet-only; no GraphQL subscriptions. Editorial #2 despite latency #4 because of GraphQL and RU pricing.
Best for: teams that want gRPC, GraphQL, and archival on Sui's provider table, with the second-best latency in this sample.
Sui's table lists gRPC (testnet and mainnet), GraphQL (mainnet), and archival (mainnet). Second in every city (39 to 70 ms). Sui methods cost 40 API credits each. Build starts at $49/month with 80M credits and 50 RPS; free access is a time-limited trial (about 10M credits, 15 RPS), not a permanent free tier.
Trade-off: no permanent free plan. At 40 credits per Sui call, Build's 80M credits cover about 2M calls. Dedicated Sui clusters are enterprise-led. Editorial #3; latency-sample #2.
Best for: multi-chain Ankr accounts that need Sui gRPC and a large Freemium credit pool.
Sui's table lists gRPC, GraphQL, and archival. Freemium: 200M API credits/month after sign-in. gRPC bills 10 credits per call plus data transfer. GraphQL and archive gRPC are Premium-only on Ankr's docs.
Trade-off: highest latency of the five in the sample (129 to 205 ms). GraphQL is not on Freemium.
Best for: products already on Alchemy that only need mainnet gRPC.
Sui's table lists Alchemy for mainnet gRPC only. Free tier: 30M CU/month at 300 CU/s (Alchemy pricing plans). Unary Sui gRPC is 10 CU for GetServiceInfo and 20 CU for most other methods; streams are $75/TB (CU costs). Third on latency in the sample (55 to 96 ms).
Trade-off: GraphQL and archival are N/A on Sui's table. Rank five is about the thinner Sui interface set, not about whether Alchemy appears in CompareNodes.
How this ranking was scored
Two ordered lists appear in this guide. Do not mix them.
| List | Order | What it optimizes for |
|---|---|---|
| Latency sample only | Tatum, QuickNode, Alchemy, Chainstack, Ankr | Lowest ms in the five-city table (Frankfurt, Virginia, Singapore, Mumbai, Tokyo). |
| Editorial production rank | Tatum, Chainstack, QuickNode, Ankr, Alchemy | gRPC now, GraphQL/archival when needed, published pricing, then latency. Chainstack is editorial #2 for GraphQL and flat RU pricing even though it is latency #4. |
Four checks decide the editorial rank. None require a CompareNodes listing or a public gRPC URL in that directory.
- gRPC now. Sui's public JSON-RPC is already off. A provider without gRPC is out.
- Interfaces for the workload. GraphQL and archival from Sui's provider table and vendor docs.
- Published pricing. Free tier or trial, paid entry price, and how Sui calls are billed.
- Multi-region latency. The five-city sample. GraphQL and flat RU pricing can outweigh a pure ms gap: Chainstack is editorial #2 and latency #4; QuickNode is editorial #3 and latency #2.
Author note: Tatum publishes this guide and the latency table. Tatum is editorial #1 and latency #1. The trade-off line under Tatum is intentional: no GraphQL in the Sui reference, and Tatum is absent from Sui's own provider table.
Sui RPC providers compared: gRPC, GraphQL, archival, pricing
Cells below are limited to what each source publishes. "Sui table" means Sui's RPC and data providers page, rechecked October 29, 2026. That page lists QuickNode, Ankr, Alchemy, and Chainstack. It does not list Tatum or BlockEden.xyz. CompareNodes lists Tatum, Chainstack, QuickNode, Ankr, and BlockEden; Alchemy is missing from that directory. Missing from a directory is not treated as a quality score here.
| Check | Tatum | Chainstack | QuickNode | Ankr | Alchemy |
|---|---|---|---|---|---|
| Editorial rank | 1 | 2 | 3 | 4 | 5 |
| Latency-sample rank | 1 | 4 | 2 | 5 | 3 |
| Best for | Lowest latency + app platform | GraphQL + flat RU | Full Sui table + next latency | Large Freemium credits | Existing Alchemy stack |
| gRPC | Yes. Mainnet, testnet, devnet hosts; seven services in docs. | Testnet and mainnet (Sui table). | Testnet and mainnet (Sui table). | Testnet and mainnet (Sui table). | Mainnet (Sui table). |
| GraphQL | Not in the Tatum Sui reference. | Mainnet (Sui table). Docs: every request is 1 RU. | Mainnet (Sui table). | Mainnet (Sui table); Premium-only on Ankr docs. | N/A (Sui table). |
| Archival or extended history | Paid plans include archive data on the Sui chain page. No public checkpoint count. | N/A on Sui table; Extended history on Chainstack docs. | Archival on mainnet (Sui table). | Archival on mainnet (Sui table); archive gRPC Premium-only. | N/A (Sui table). |
| Pricing and free tier | Free: 100K lifetime credits, 3 RPS. Starter from $49/mo, 4M credits, 200 RPS. | Free: 3M RU/mo, 25 RPS. Growth $49/mo, 20M RU, 250 RPS. 1 RU per call. | Trial (~10M credits, 15 RPS). Build $49/mo, 80M credits, 50 RPS; Sui = 40 credits/call. | Freemium: 200M credits/mo. gRPC 10 credits/call + transfer. Premium for GraphQL. | Free: 30M CU/mo, 300 CU/s. Unary 10 to 20 CU; streams $75/TB (Alchemy CU docs). |
| On Sui's provider table | No | Yes | Yes | Yes | Yes (gRPC only) |
| Networks on CompareNodes (context) | Devnet, mainnet, testnet | Mainnet, testnet | Mainnet, testnet | Mainnet, testnet | Not listed (directory gap, not a quality score) |
| Latency in the five-city sample | Lowest in every city (18 to 33 ms) | Fourth in every city (72 to 112 ms) | Second in every city (39 to 70 ms) | Highest in every city (129 to 205 ms) | Third in every city (55 to 96 ms) |
| Trade-off | No GraphQL in Sui docs; not on Sui's table. | Fourth on latency; GraphQL mainnet-only. | No permanent free tier; 40 credits per Sui call. | Highest latency; GraphQL Premium-only. | No GraphQL or archival on Sui's table. |
What CompareNodes shows for Sui
CompareNodes listed 60 Sui providers on the October 29, 2026 check. Three of them list devnet, mainnet, and testnet: Tatum, RedSwitches, and BlockEden.xyz. That is directory coverage, not a latency or GraphQL score. BlockEden is covered below as an expected alternative outside the top five.
CompareNodes also publishes public gRPC URLs for Tatum (useful for smoke tests without an account):
- Mainnet:
grpcs://sui-mainnet-grpc.gateway.tatum.io:443 - Testnet:
grpcs://sui-testnet-grpc.gateway.tatum.io:443
Most commercial peers keep Sui endpoints behind an account. That is normal. It is not a ranking criterion in this guide.
Sui networks listed on CompareNodes
Directory checked October 29, 2026. A count of 3 means devnet, mainnet, and testnet. Context only.
Source: comparenodes.com/protocols/sui. RedSwitches also lists 3; omitted from the bars for space.
Sui RPC latency by city
Round-trip Sui latency, one city at a time. Numbers come from the Sui table in Tatum's RPC benchmark (published August 28, 2025). That post attributes the figures to independent CompareNodes-style global benchmarks and points readers to chaintester.io for public retests. Sample: five cities, five providers, one ms figure per cell. No per-city sample size or raw request log is published in that article. Figures were re-read against that page on October 29, 2026. The fastest number gets a mint underline. Tatum is that number in every city. QuickNode is next in every city.
Sui latency by city
Pick a city. Bars are round-trip milliseconds. Lower is faster.
18 ms Tatum is lowest in Frankfurt.
20 ms Tatum is lowest in Virginia.
29 ms Tatum is lowest in Singapore.
30 ms Tatum is lowest in Mumbai.
33 ms Tatum is lowest in Tokyo.
Source: Sui latency table in Tatum's RPC benchmark (Aug 28, 2025), rechecked Oct 29, 2026. Same five cities, same five providers. Method details above.
Live check on October 29, 2026: sui_getLatestCheckpointSequenceNumber against https://fullnode.mainnet.sui.io returned method not found, with the message that JSON-RPC on public full nodes has been deprecated and to migrate to gRPC or GraphQL. That matches Sui's migration guide. Tatum's JSON-RPC gateway still documents the same method for clients that have not finished migrating; production calls need an API key.
Which provider fits the workload
| Workload | Start with | Also look at |
|---|---|---|
| Wallet, game, or consumer app | Tatum. gRPC streaming, dedicated rate options, and webhooks for Sui events. | QuickNode if the screen already uses their GraphQL endpoint. |
| App that also runs on other chains | Tatum. The same platform covers 142 networks, including the Sui gateway. | Alchemy if the rest of the stack is already there and Sui stays on mainnet gRPC. |
| Lowest latency in the five-city sample | Tatum. 18 to 33 ms across Frankfurt, Virginia, Singapore, Mumbai, and Tokyo. | QuickNode is next in that sample, at 39 to 70 ms. |
| GraphQL and flat request pricing | Chainstack. Mainnet GraphQL; free 3M RU/month; Growth at $49/month. | QuickNode if you already pay for their credit pool and need archival on the Sui table. |
| Large free credit pool for tests | Ankr Freemium (200M credits/month). GraphQL and archive gRPC need Premium. | Alchemy's 30M CU free tier for mainnet gRPC-only experiments. |
1. Tatum: gRPC, pricing, and trade-offs
Editorial rank: 1. Latency-sample rank: 1 (18 to 33 ms). Best for: production apps that need gRPC, a JSON-RPC gateway during migration, and multi-region latency.
Interfaces: The Sui RPC reference documents JSON-RPC and gRPC on mainnet, testnet, and devnet. gRPC services include LedgerService, StateService, MovePackageService, NameService, SignatureVerificationService, SubscriptionService (SubscribeCheckpoints), and TransactionExecutionService. GraphQL: not in the Tatum Sui reference. Archival: paid plans include archive data on the Sui chain page; no public checkpoint count. On Sui's provider table: no.
Pricing (checked October 29, 2026): Free: 100K lifetime credits, 3 RPS, 2 API keys (plans and limits). Starter from $49/month: 4M credits/month, 200 RPS. Dedicated API keys from $99/month for 10 reserved RPS (do not consume plan credits). Auth: x-api-key. Platform coverage: 142 networks on one account.
Trade-off: No GraphQL in the Sui reference. Not listed on Sui's RPC and data providers table (October 29, 2026). Latency evidence is Tatum-published.
2. Chainstack: GraphQL, flat RU pricing, and trade-offs
Editorial rank: 2. Latency-sample rank: 4 (72 to 112 ms). Best for: mainnet GraphQL with flat request units and a path to dedicated or self-hosted Sui nodes.
Interfaces: Sui's table: gRPC (testnet and mainnet), GraphQL (mainnet), archival N/A. Chainstack docs: JSON-RPC, WebSocket, and gRPC on the same node; every request is 1 RU, including GraphQL (Chainstack Sui guide, updated October 12, 2026). Extended history, not genesis archive.
Pricing (checked October 29, 2026): Free Developer: 3M RU/month, 25 RPS. Growth: $49/month, 20M RU, 250 RPS. Pro: $199/month, 80M RU, 400 RPS.
Trade-off: Fourth on latency in every city versus Tatum 18 to 33 ms and QuickNode 39 to 70 ms. GraphQL is mainnet-only; no GraphQL subscriptions. Editorial #2 for GraphQL and RU pricing despite latency #4.
3. QuickNode: gRPC, GraphQL, archival, and pricing
Editorial rank: 3. Latency-sample rank: 2 (39 to 70 ms). Best for: teams that want gRPC, GraphQL, and archival on Sui's provider table with the second-best latency in this sample.
Interfaces: Sui's table: gRPC (testnet and mainnet), GraphQL (mainnet), archival (mainnet). CompareNodes: mainnet and testnet. Archive reads on mainnet route automatically per QuickNode Sui docs. Sui methods cost 40 API credits each.
Pricing (checked October 29, 2026): Time-limited free trial (~10M credits, 15 RPS, one endpoint). Build from $49/month: 80M credits, 50 RPS. At 40 credits per Sui call, Build covers about 2M Sui calls before overage.
Trade-off: No permanent free tier. At the same $49 entry, Chainstack Growth offers 20M RU at 1 RU/call versus ~2M Sui calls on QuickNode Build. Dedicated Sui clusters are enterprise-led. Editorial #3; latency-sample #2.
4. Ankr: gRPC, GraphQL, free credits, and trade-offs
Editorial rank: 4. Latency-sample rank: 5 (129 to 205 ms). Best for: multi-chain Ankr accounts that need Sui gRPC and a large Freemium credit pool.
Interfaces: Sui's table: gRPC, GraphQL, archival. Ankr docs: GraphQL and archive gRPC are Premium-only; Freemium still reaches gRPC (Ankr Sui).
Pricing (checked October 29, 2026): Freemium: 200M API credits/month after sign-in. gRPC: 10 credits per method call plus 5,000,000 credits per GB transfer (~$0.50/GB). Premium pay-as-you-go from about $10 per 100M credits (service plans).
Trade-off: Highest latency of the five in the sample. GraphQL is not on Freemium. Dedicated Sui nodes are sales-led in public materials.
5. Alchemy: mainnet gRPC, pricing, and trade-offs
Editorial rank: 5. Latency-sample rank: 3 (55 to 96 ms). Best for: products already on Alchemy that only need mainnet gRPC.
Interfaces: Sui's table: mainnet gRPC only; GraphQL N/A; archival N/A. Alchemy documents all seven gRPC services. CompareNodes absence is a directory gap, not a quality score.
Pricing (checked October 29, 2026): Free: 30M CU/month, 300 CU/s (pricing plans). Unary Sui gRPC: GetServiceInfo 10 CU; most other unary methods 20 CU. Streaming methods: $75/TB delivered (compute unit costs). PAYG listed around $0.52 to $0.525 per 1M CU on Alchemy pricing pages.
Trade-off: No GraphQL or archival on Sui's table. Faster than Chainstack and Ankr in the sample, but thinner Sui interface set than QuickNode or Chainstack. No documented dedicated Sui cluster in Alchemy's Sui docs.
Also consider: BlockEden.xyz
Editorial rank: outside top five. Latency-sample rank: not in the five-provider table. Best for: a Sui JSON-RPC marketplace key with a published free CU pool and three networks on CompareNodes.
Interfaces: CompareNodes: devnet, mainnet, and testnet (with Tatum and RedSwitches). Public materials emphasize JSON-RPC at 200 CU/request. Not on Sui's provider table (October 29, 2026).
Pricing (checked October 29, 2026): Free: 10M CU/day, 1 RPS, 15 projects. Basic: $19.99/month, 40M CU/day, 10 RPS. Pro: $49.99/month, 100M CU/day (BlockEden pricing).
Trade-off: Confirm gRPC and GraphQL directly before treating BlockEden as a drop-in for the Sui migration stack. Not in the latency sample used above.
Read a Sui checkpoint on Tatum
Production calls send x-api-key. The method below is the connectivity check while JSON-RPC is still on the gateway. New code should call gRPC LedgerService/GetCheckpoint, as in the Sui gRPC reference.
curl --request POST \
--url https://sui-mainnet.gateway.tatum.io \
--header 'content-type: application/json' \
--header 'x-api-key: YOUR_API_KEY' \
--data '{
"jsonrpc": "2.0",
"id": 1,
"method": "sui_getLatestCheckpointSequenceNumber",
"params": []
}'grpcurl \
-H "x-api-key: YOUR_API_KEY" \
sui-mainnet-grpc.gateway.tatum.io:443 \
sui.rpc.v2.LedgerService/GetServiceInfoFor a longer walkthrough of objects, checkpoints, and the gateway URL, see Building on Sui.
FAQ
Tatum, for production apps that need gRPC and multi-region latency. It is lowest in all five cities against QuickNode, Alchemy, Chainstack, and Ankr. Trade-off: no GraphQL in the Sui reference, and it is not on Sui's provider table.
Editorial: Tatum, Chainstack, QuickNode, Ankr, Alchemy. Latency sample only: Tatum, QuickNode, Alchemy, Chainstack, Ankr. BlockEden.xyz is outside the top five but lists all three networks on CompareNodes.
Tatum. Latency is lower in Frankfurt, Virginia, Singapore, Mumbai, and Tokyo. In Frankfurt it is 18 ms versus 39 ms. QuickNode remains the next-fastest peer in that sample.
Yes. Mainnet is sui-mainnet-grpc.gateway.tatum.io:443, with matching testnet and devnet hosts. Services cover checkpoints, objects, streaming, execution, and simulation.
On Sui's provider table: QuickNode, Ankr, and Chainstack list GraphQL on mainnet. Alchemy and Tatum do not. Ankr's docs mark GraphQL as Premium-only.
Sources
- CompareNodes, Sui providers, directory checked October 29, 2026 (60 providers).
- Tatum RPC benchmark, Sui five-city latency table, published August 28, 2025, rechecked October 29, 2026.
- Sui JSON-RPC migration guide, July 27 and mid-October 2026 timeline.
- Sui RPC and data providers, rechecked October 29, 2026 (includes Chainstack; does not list Tatum or BlockEden).
- Tatum Sui RPC reference, Sui gRPC reference, and plans and limits.
- QuickNode pricing and Sui API credits (40 credits per call).
- Chainstack Sui provider guide for published RU tiers and the 1 RU per GraphQL request note (updated October 12, 2026).
- Ankr service plans and Ankr Sui docs (GraphQL Premium-only).
- BlockEden pricing and Sui marketplace (200 CU per JSON-RPC request).
