7 Best Sui RPC Providers in 2026: Pricing & Features Compared

What matters for Sui RPC in 2026

Sui's API transition makes a generic “supports Sui” checkbox insufficient. The Sui Foundation disabled JSON-RPC on its mainnet full nodes during the week of July 27, 2026; removal from the codebase is planned for mid-October. Its documented replacements are gRPC and GraphQL. Commercial providers can have different legacy schedules. Source: Sui API reference.

For a new integration, start with the replacement interface your SDK and workload require. Check streaming separately from ordinary reads. For an existing application, inventory every JSON-RPC method before changing a URL: request shapes, pagination, authentication, and response handling may also change.

Treat historical data as a separate requirement. An endpoint that reads a recent object may not retrieve an old object version or search months of events. Ask for the retention policy of each service rather than accepting a broad “archive” label. Build your migration test around real application queries, including a query that should return no result, so an empty response is not mistaken for complete coverage.

Sui RPC providers: pricing and features at a glance

Prices below are public monthly starting points checked on October 9, 2026, in USD. Promotional and annual discounts are excluded. Plan allowances use different billing units; they are not equivalent request counts. “Not confirmed” means the sources reviewed did not establish the feature, rather than that it is unavailable.

Interface details: OnFinality Sui, QuickNode Sui, Chainstack GraphQL, Ankr Sui, and BlockVision gRPC.

Use this table to eliminate incompatible candidates, then compare a common workload. A million lightweight calls and a million large historical responses can produce very different bills. Trial allowances also do not establish the capacity available on a paid plan.

1. OnFinality — managed Sui gRPC with room to scale

Best fit to evaluate: teams that want managed Sui access while keeping a path to dedicated infrastructure.

OnFinality documents a TLS gRPC endpoint with API-key authentication for Sui, including reads, transaction execution, and streaming. Its network page lists Virginia and Hong Kong regions. GraphQL support is not confirmed in the sources used for this comparison. Source: OnFinality Sui.

Growth includes a 200 RU/second limit and $6 per million RU overage. Response units account for request complexity and response size, so do not read the allowance as a fixed number of calls. Dedicated nodes have separate pricing. Source: RPC plans.

The reason to shortlist OnFinality is operational fit: start by checking whether your application can use its documented interface, then evaluate how comfortably the shared plan handles your request mix. A team already using OnFinality on another network can assess whether one operating workflow simplifies endpoint management.

Tradeoff: if your frontend depends on GraphQL filtering, establish that requirement before choosing a gRPC endpoint. Ask about historical coverage for your exact methods; this guide does not claim genesis-to-present access for every query. OnFinality publishes this article and appears first as the publisher's offering, not as the winner of a latency benchmark.

2. Chainstack — gRPC and GraphQL for an existing node workflow

Best fit to evaluate: teams already operating networks through Chainstack or needing its Sui GraphQL service.

Chainstack documents a mainnet GraphQL endpoint alongside gRPC. Its GraphQL retention varies by query: filtered transactions and events have a 90-day window, while point lookups follow the underlying node's retention. This is not an unlimited historical-search promise. Sources: GraphQL endpoint, Sui GraphQL and gRPC.

Growth publishes a 250 RPS limit and $15 per million overage; the free plan lists 25 RPS. Source: Chainstack pricing.

The attractive case is workflow continuity. If your team already understands Chainstack's provisioning and authentication, testing Sui through the same workflow may reduce integration overhead. That advantage should be weighed against the data your application actually needs, particularly for historical analytics.

Tradeoff: GraphQL availability does not imply every query covers all chain history. Write down your oldest required timestamp and test both point retrieval and filtered search at that boundary. Treat dedicated or unlimited-node products as separate purchases rather than assuming that a shared-plan subscription includes isolated resources.

3. Dwellir — response-based pricing for Sui access

Best fit to evaluate: developers who prefer a response allowance when estimating a shared RPC budget.

Dwellir documents Sui gRPC and legacy JSON-RPC. Its Developer plan allows 100 responses per second and charges $5 per million responses beyond the included allowance. The documentation describes one response as one API credit on shared plans. Source: Dwellir Sui documentation.

This pricing model is useful when a team has a reasonably stable query mix and wants to relate its logs to the provider's usage dashboard. It still deserves a workload test: a chatty application can exhaust an allowance through polling even when each response is small.

Tradeoff: a response allowance alone says little about streaming behavior, historical retention, or how a burst is handled. Confirm the limits for persistent connections and the oldest data your queries can retrieve. GraphQL was not established by the sources reviewed, so teams requiring it should obtain confirmation before including Dwellir in that shortlist. Keep migration compatibility and long-term interface support as separate checklist items.

4. QuickNode — documented gRPC and GraphQL on mainnet

Best fit to evaluate: Sui applications combining direct chain interaction with GraphQL queries.

QuickNode documents mainnet gRPC, GraphQL, and legacy JSON-RPC. Its Sui archive access is available through gRPC in the US region; its testnet matrix does not include GraphQL or archive. Source: QuickNode Sui documentation.

Build lists 50 requests per second alongside the credit allowance. Monthly billing is $49; the lower displayed annual rate is a different commitment. Sui methods have their own credit schedule. Sources: pricing, Sui API credits.

For a team needing both interfaces, the practical benefit is being able to evaluate them within one provider. Reproduce your application's object reads, transaction submission, and GraphQL queries during the trial instead of judging the service from a single successful call.

Tradeoff: testnet support is not identical to mainnet support. A US archive endpoint may also behave differently from a regional live endpoint for users elsewhere. Model method weights before comparing 80 million credits with another provider's response allowance. The larger number is not automatically the larger usable budget.

5. Ankr — usage-based gRPC pricing with bandwidth charges

Best fit to evaluate: applications with variable call volume that can track transferred data.

Ankr documents Sui gRPC on mainnet and testnet, premium archival gRPC, and premium mainnet GraphQL. Its Sui documentation flags the JSON-RPC migration. Source: Ankr Sui APIs.

The published gRPC rate uses 10 API credits per call at $0.10 per million credits, equivalent to $1 per million calls, plus $0.50 per GB transferred in both directions. These gRPC rules differ from its legacy non-EVM HTTPS weighting. Source: Ankr pricing.

This can be a useful candidate for uneven traffic, but the call price is only part of the budget. Capture the request and response sizes of your real workload, including subscriptions and historical reads, before forecasting spend.

Tradeoff: do not estimate a streaming application's cost from a tiny unary query. Confirm which premium services your key unlocks and whether each interface meets your retention requirements. Compare the resulting monthly total with subscription plans after accounting for bandwidth, rather than selecting a provider from its headline per-call figure alone.

6. GetBlock — several deployment options for Sui workloads

Best fit to evaluate: teams comparing shared access against isolated capacity or a dedicated node.

GetBlock's Sui page lists gRPC and legacy JSON-RPC for shared access. It also lists isolated Limitless tiers from $150 per month and dedicated Sui nodes from $1,700 per month; GraphQL is presented through an indexer product. Source: GetBlock Sui.

That range makes GetBlock relevant when the procurement question is broader than the lowest shared-plan price. Evaluate whether your bottleneck is a monthly allowance, peak throughput, historical access, or contention, then request the product matching that constraint.

Tradeoff: shared, isolated, dedicated, and indexed services have different scopes. Confirm compute-unit weights, supported methods, and archive configuration for the specific tier you intend to buy. Do not treat the existence of a GraphQL indexer as proof that the standard shared endpoint implements every Sui GraphQL query. Compare an explicit quote with your workload assumptions before making a deployment decision.

7. BlockVision — gRPC alongside indexed application data

Best fit to evaluate: applications needing Sui RPC plus a separate source of enriched or indexed data.

BlockVision documents Sui gRPC on paid Lite and higher tiers. Lite includes 500 CU per second; at the documented 50 CU per Sui call, its 100 million monthly CU corresponds to two million calls and its throughput to ten calls per second. The free tier does not include Sui gRPC. Sources: gRPC reference, billing and limits.

Its indexing API is a distinct product with its own methods and requirements. Source: Sui indexing API.

This combination is worth evaluating if application screens need data that would otherwise require you to build an indexer. Measure how many RPC calls an indexed result could replace, while checking that the data model matches your product.

Tradeoff: the apparent size of a compute-unit allowance can hide a much smaller call allowance. Also separate transaction-critical operations from enriched data reads: an indexing response has different freshness and completeness requirements. Validate both paths instead of treating them as interchangeable.

How to choose the best Sui RPC provider for your application

Begin with three concrete requirements: the interface your SDK uses, the oldest data you must access, and your busiest minute of traffic. These usually eliminate more unsuitable plans than a list of general platform features.

For a wallet or transaction backend, prioritize the path from object reads and transaction preparation to execution and confirmation. Test errors as well as successful transactions. A response timeout leaves uncertainty about execution; your application should reconcile the transaction status before retrying a state-changing operation.

For a DeFi application, replay periods with many active users and large response payloads. Check whether a request limit remains comfortable when the page refreshes several balances or objects at once. Put a budget on the entire user interaction rather than on one RPC method.

For an analytics or indexing system, create a query matrix covering recent data, older checkpoints, old object versions, and filtered event searches. Mark which answers must come from the node and which can come from your own database. A service that returns current state correctly may still be unsuitable for a backfill.

Run a small, controlled comparison from your deployment region. Use the same request set, connection strategy, concurrency, and time windows on each candidate. Track median and p95 latency, error rate, checkpoint lag, reconnect behavior, and usage charged. Increase concurrency gradually within each plan's allowed limits. Keep cold-start and warm-connection results separate.

Finally, test recovery. Check whether your second endpoint supports the same interface and methods, and whether your application can resume streaming without silently losing data. Independent failure recovery can matter more than a minor improvement in median response time. None of the providers here has been assigned a measured performance score by this article.

Which Sui RPC provider should you shortlist?

Choose OnFinality or Dwellir as candidates when managed gRPC and a clear shared-plan budget match your needs. Compare QuickNode, Chainstack, and Ankr when mainnet GraphQL is a firm requirement, paying particular attention to retention and access tier. Evaluate GetBlock when resource isolation is part of the decision. Consider BlockVision when indexed application data could reduce the work of maintaining your own data layer.

The right choice is the provider that passes your API, history, throughput, and recovery tests at an acceptable total cost. Start with OnFinality's Sui network details and RPC pricing if that operating model fits, or return to the RPC comparison hub to explore other networks.

Frequently asked questions