Base RPC Providers Compared: Pricing, Flashblocks Support and Denim Readiness

Base RPC Providers Compared: Features, Pricing and Denim Readiness

Base’s planned Denim upgrade is set to replace Flashblocks with canonical blocks and RPC streams every 200 milliseconds. That could change how applications receive Base’s fastest transaction updates, making providers’ migration plans as important as their current feature lists.

• Chainstack’s comparison recommends Chainstack, but it was written by the provider, not based on an independent performance ranking.
• Billing units and method costs vary, so headline allowances are not directly comparable.
• Five providers document Flashblocks support, though the features they list differ.
• Ask providers how they plan to handle Denim before choosing a production endpoint.

Why Base RPC choice matters

An RPC provider lets an application request blockchain data and submit transactions. If that connection is slow, limited or costly, users may notice, even when the network is working normally.

Base currently produces a block every two seconds. Flashblocks provide sequencer preconfirmations every 200 milliseconds. A preconfirmation is an early indication that a transaction has been accepted for inclusion. It is not the same as a sealed, canonical block. Some providers make this early data available through requests using the pending tag or through WebSocket subscriptions, persistent connections that stream updates to an application.

Base’s upgrade schedule lists Denim for Base Sepolia on October 21, 2026, at 18:00 UTC, with Base Mainnet targeting November 2026. The upgrade is scheduled, but has not been confirmed as activated. Base says Denim will introduce native blocks every 200 milliseconds and replace Flashblocks with canonical block and RPC streams. In plain terms, faster updates are expected to become part of the regular block and RPC flow instead of relying on the current Flashblocks mechanism.

Low-latency access will still matter. Applications may just need different ways to receive it. The provider comparison does not explain how any of the six services will handle the transition, or whether existing pending-tag calls and subscriptions will change.

Base’s software and its compatibility are also separate questions. A Chainstack-authored guide says Base moved off OP Stack node software after the Azul upgrade while remaining compatible with OP Stack specifications. That distinction can matter when checking node-method compatibility, but teams should verify it against current Base technical documentation.

What this comparison can and cannot tell you

The provider details below come from a Chainstack-authored comparison of Base RPC providers for onchain applications. It recommends Chainstack, so its commercial interest matters when assessing its conclusions. The comparison says it checked provider documentation and pricing pages on October 20, 2026. Its detailed claims have not been independently verified here. Treat them as a dated guide to listed features and plan terms, not as an audit or performance test. Confirm current documentation, pricing and plan limits before deciding.

A feature in the documentation shows what a provider says it supports. It does not establish latency, uptime, geographic coverage or support quality. And if a feature is missing from the documentation reviewed, that does not prove the provider cannot offer it.

Six Base RPC providers compared

Chainstack: broad infrastructure options

The comparison lists Base Mainnet and Sepolia access through Chainstack Global Nodes, Dedicated Nodes and Chainstack Self-Hosted, along with support for more than 70 networks. It says Global Nodes provide full and archive data, WebSocket access, and debug and trace methods. Archive access lets you query historical chain data. The comparison does not specify the exact retention depth or coverage.

For Flashblocks, Chainstack documents pending-tag access, eth_sendRawTransactionSync, and WebSocket subscriptions such as newFlashblocks, newFlashblockTransactions and pendingLogs.

The listed pricing charges one request unit for a full call and two for an archive call. Calls less than 127 blocks behind the tip count as full. The comparison lists a $49 Growth plan with 20 million request units and a limit of 250 requests per second. Archive data and debug and trace calls require a paid plan. Debug and trace calls cost two units. The free plan limits eth_getLogs to 100 blocks per request. The comparison also says Chainstack’s Base nodes do not expose optimism_* methods.

Chainstack stands out in the listed options for teams that want self-hosted or dedicated infrastructure. That is a distinction based on features, not proof that it will perform best for every application.

Quicknode: wide documented coverage, method-weighted costs

The comparison lists Quicknode support for Base archive data, WebSockets, debug and trace APIs, Flashblocks and OP-Node APIs. It says Flashblocks works on existing Base endpoints on Mainnet and Sepolia without an add-on. Quicknode lists SOC 2 Type 2 and ISO/IEC 27001 credentials. Dedicated clusters are available on its Enterprise plan.

Pricing is based on the method, not a flat charge for every Base request. The comparison gives eth_call as an example at 20 credits and says debug and trace methods cost double. It lists a $49 Build plan with 80 million credits. At 20 credits per call, that allowance would cover about four million calls of that type, before other methods or plan conditions are taken into account. The comparison says Quicknode offers a one-month trial rather than a permanent free tier, and does not list a self-hosted option.

Quicknode may suit teams that need the documented API range and listed compliance credentials. The cost depends on which methods the application actually uses.

Dwellir: per-response billing and streamed data

The comparison lists Dwellir Base Mainnet and Sepolia access over HTTPS and WebSocket, with archive access and trace and debug APIs on paid plans. Its Flashblocks documentation includes pending-tag access, eth_sendRawTransactionSync, and subscriptions such as newFlashblocks, newFlashblockTransactions and pendingLogs.

The listed billing model charges one credit per response for every method, including debug and trace, and one credit for each WebSocket notification. A notification is an individual update sent over the persistent connection. The comparison lists a $49 Developer plan with 25 million responses and a 100-requests-per-second limit. Additional responses cost $5 per million. Dedicated Base clusters are listed from $3, 366 a month.

On the listed Developer plan, eth_getLogs is limited to 500 blocks per request. The comparison says the method is unavailable on the free plan. Per-response pricing may be straightforward to model, but frequent streamed notifications also count. Confirm archive coverage and current plan limits.

Ankr: a large listed free allowance, with narrower documented features

Ankr’s listed Base support includes selected Flashblocks-related reads through the pending tag, including eth_getBlockByNumber, eth_getBalance and eth_getTransactionCount, as well as preconfirmed transactions and receipts. According to the comparison, Ankr’s Base method reference does not list Flashblocks subscriptions or debug and trace methods.

The comparison lists 200 million credits a month on Freemium and Premium pricing of $0.10 per million credits. It says an EVM call costs 200 credits, which would put one million such calls at about $20 at the stated rate. That estimate applies to calls billed at that rate, not every method or workload. WebSocket access is listed as a Premium feature.

Dedicated capacity is described as customer-tailored through Enterprise. The comparison attributes a SOC 2 Type II attestation to an infrastructure operator, rather than making an unqualified claim about every Ankr service. Ask which entity and services the report covers.

Alchemy: debug support and dedicated clusters

The comparison lists Alchemy support for Base over HTTPS and WebSocket, a Debug API, and Flashblocks access through the pending tag and subscriptions for preconfirmed data at roughly 200-millisecond intervals. It lists single-tenant Dedicated Clusters for Base and says Alchemy reports SOC 2 Type II certification. Check the provider’s current materials to confirm which entity and services are covered.

The comparison says Alchemy’s Base page lists no Trace API. Its compute-unit costs vary by method: it gives 26 units for eth_call and 60 for eth_getLogs. At 60 units per log query, 30 million free units would cover about 500, 000 queries of that type, before other usage is counted. The comparison does not list a self-hosted option or ISO 27001 credential.

Alchemy may suit teams that need debug access or dedicated infrastructure but do not require a documented Trace API. Check current method costs and free-plan terms before estimating spend.

dRPC: requests routed across providers

dRPC routes requests across independent node providers and supports Base over HTTPS and WebSocket. The comparison lists paid usage at $6 per million requests, with archive requests priced the same as full requests and throughput of up to 5, 000 requests per second.

Its paid tier lists four debug methods: debug_traceBlockByHash, debug_traceBlockByNumber, debug_traceTransaction and debug_traceCall. The comparison does not list Flashblocks or trace_* methods in dRPC’s Base documentation. It says the free tier uses public nodes, allows up to 100 requests per second, and does not include eth_getLogs or filters.

Routing requests across providers offers a different approach from relying on a single node operator, but the provider serving a request may affect its result. The comparison does not list a dedicated-node tier or SOC 2 or ISO 27001 credentials for dRPC.

Compare costs using your own workload

Request units, credits, compute units and responses are provider-specific billing measures. A million units from one service are not necessarily equivalent to a million from another. Even with one provider, a basic read, a log query, an archive request and a trace operation can have different costs.

Estimate a representative month of usage before choosing a plan. Include ordinary reads, eth_getLogs, historical queries, WebSocket notifications, retries, and debug or trace calls. Then apply each provider’s current method-level pricing and rate limits to the same workload.

A listed feature is not the same as a measured result. The comparison does not provide a reproducible, provider-by-provider benchmark with disclosed test methods and results, so it cannot show which service is fastest or most reliable for a particular application.

Key questions about Base RPC providers

• Which provider should a Base team choose?

  The available comparison does not establish an independent winner. Teams looking for self-hosted or dedicated infrastructure may investigate Chainstack. Those focused on method-level API coverage may compare Quicknode or Alchemy. Teams considering routed access, per-response billing or a large listed free allowance may assess dRPC, Dwellir or Ankr. Test providers against your own workload.

• Which providers document Flashblocks support?

  The comparison lists Chainstack, Ankr, Dwellir, Alchemy and Quicknode. Ankr’s listed support is narrower: selected pending reads and preconfirmed data, with no Flashblocks subscriptions listed in its Base method reference.

• When is Denim scheduled?

  Base Sepolia is listed for October 21, 2026, at 18:00 UTC, and Base Mainnet is targeting November 2026. These are schedule dates, not confirmation that the upgrade has been activated.

• Does a feature list prove a provider is faster or more reliable?

  No. Documentation shows listed support, not measured performance or contractual guarantees. Ask for relevant service-level terms, then test the methods and regions your application will use.

Choose for the workload and plan for Denim

Chainstack’s comparison recommends its own service for most teams, citing its Base features, infrastructure options and request-unit pricing. That is a vendor’s assessment, not an independent ranking. The useful lesson is to match a provider’s documented features and billing rules to the application you actually run.

Before committing, verify current plan terms, clarify what any compliance report covers, and ask how the provider will handle Denim. Find out whether pending-tag calls and existing WebSocket subscriptions will continue, change or be replaced, and whether your application will need updates. A feature table can help narrow the choices. It cannot make the migration decision for you.