Top 7 Ethereum RPC providers for custodians and asset managers in 2026

Created Aug 17, 2026 Updated Oct 29, 2026

Chainstack

in Compare

Ethereum is the settlement layer institutional custody platforms build against — the chain where tokenized real-world assets, stablecoin reserves, and regulated fund positions actually settle. On March 17, 2026, the SEC and CFTC jointly classified ETH as a digital commodity under federal guidance, closing more than a decade of security-versus-commodity ambiguity and setting the compliance baseline that qualified custodians and asset managers now design infrastructure around.

The network backs that role at scale: Ethereum settles roughly 2.6 million transactions a day as of August 2026, and secures $41.8 billion in DeFi TVL — an increasing share of it custodied on behalf of institutions rather than held in retail wallets. The compliance clock is running in parallel: the EU’s MiCA regime and the US GENIUS Act both push custodians toward auditable, SLA-backed infrastructure instead of best-effort public endpoints, a shift covered in Chainstack’s 2026 crypto regulation guide.

This guide compares seven Ethereum RPC providers specifically through the lens of custody and asset management: archive depth, audit-trail methods, published compliance certifications, and the throughput and support commitments that regulated infrastructure teams actually sign off on.

Quick verdict: best Ethereum RPC providers for custodians and asset managers in 2026

Chainstack is our top pick for Ethereum custody and asset-management workloads (an editorial judgment, not a benchmark result), based on full archive with debug and trace at flat 1–2 RU pricing, published SOC 2 Type II and ISO 27001 certification, MEV protection on by default and a contractual SLA on the Enterprise plan. Chainstack publishes this page, runs the benchmark cited below and is one of the 7 listed providers, so see how we compared and the benchmark limits before relying on the ranking. In that benchmark Chainstack had the lowest overall P50, P95 and P99 of the three listed providers measured, but its US P95 trailed Alchemy and Quicknode. Last updated October 29, 2026. The best fit by need:

Best for production custody that needs compliance and cost predictability together: Chainstack (full archive and debug/trace from Growth, both certifications published, flat 1–2 RU pricing, MEV protection on by default).
Best for large institutions with a formal vendor-onboarding process: Blockdaemon (custody, staking and RPC under one contract, both certifications published; sales-led evaluation).
Best for dual-certified infrastructure with self-serve signup: Quicknode (SOC 2 Type II and ISO 27001, archive on every paid plan; method-weighted credits; fastest Germany P95 of the three providers measured in the October 2026 benchmark, 40 ms).
Best for developer tooling and a permanent free tier: Alchemy (30M CU/month free; no published ISO 27001; fastest US P95 of the three providers measured, 80 ms).
Best for evaluation and pre-production on a large free allowance: Ankr (200M API credits/month on Freemium; ISO 27001 not yet completed).
Best as failover on top of a certified primary provider: RouteMesh (routing layer across multiple backends; no independently published certification).
Best for enriched token and portfolio APIs: Uniblock (Unified APIs and pooled rate-limit headroom across 55+ providers; no independently published certification).

Provider comparison: 7 Ethereum RPC providers for custodians and asset managers (2026)

The table below summarizes public positioning as of August 2026. The rank is the editorial order used in this article (see how we compared), not a benchmark result. The last column shows the overall P95 from Chainstack’s October 2026 Ethereum benchmark, which measured only three of the seven providers (lower is better). To keep the comparison focused on custody and asset-management fit, it narrows to providers with either institutional-grade credentials or genuine product differentiation for this workload — general-purpose providers without both SOC 2 Type II and ISO 27001 published (Infura, dRPC, and GetBlock, which publishes SOC 2 Type II but no ISO 27001) are not among the 7 listed providers, though they remain reasonable choices for non-regulated Ethereum workloads.

Rank	Provider	Best for	Pricing model	Archive, debug & trace	Compliance certifications	Dedicated infrastructure	Why it matters for custodians & asset managers	Overall P95 in Oct 2026 benchmark (Chainstack-run)
1	Chainstack	Flat-priced archive/trace with both certifications and MEV protection by default	Flat RU (1 full, 2 archive/debug/trace)	Full archive + debug + trace (Growth plan+)	SOC 2 Type II (Dec 2025) + ISO 27001 (Jul 2026)	Yes (Pro plan+)	Predictable audit-workload cost, both certifications published, MEV protection on by default	78 ms
2	RouteMesh	Failover layer on top of a certified primary provider	Per-request routing fee	Inherits from upstream provider	Not independently published	Not applicable (routing layer)	Multi-provider failover for continuous wallet-monitoring uptime	Not measured
3	Uniblock	Enriched token/portfolio APIs and pooled rate-limit headroom	Per-request across 55+ providers	Inherits from upstream provider	Not independently published	Not applicable (routing layer)	Pooled rate-limit headroom plus unified token/portfolio APIs	Not measured
4	Blockdaemon	Large institutions buying custody, staking and RPC under one contract	Custom, institutional	Available (custom scoping)	SOC 2 Type II + ISO 27001	Yes	Built specifically for banks and custodians; staking and custody bundled	Not measured
5	Ankr	Evaluation and pre-production on a large free credit allowance	API credits	Archive on paid plans	SOC 2 Type II only	Available (Premium+)	Largest free credit allowance listed (200M API credits/month; credit units differ between providers); ISO 27001 not yet completed	Not measured
6	Alchemy	Developer tooling and a permanent free tier	Method-weighted CU	Archive + debug on paid plans	SOC 2 Type II only	No standard dedicated tier	Strong tooling, but no ISO 27001 blocks some dual-certification mandates	103 ms
7	Quicknode	Dual-certified infrastructure with self-serve signup	Method-weighted credits	Archive on all paid plans	SOC 2 Type II + ISO 27001	Enterprise only	Dual-certified, but pricier at scale for trace-heavy audit workloads	92 ms

⚠️ Note on compliance certifications: Only published, independently audited SOC 2 Type II and ISO 27001 reports are listed as held. Self-reported “in progress” status and undocumented assertions are not counted as certified.

How we compared Ethereum RPC providers for custodians and asset managers

Disclosure: this comparison is published by Chainstack. Chainstack is one of the 7 listed providers and is ranked first, so read the ranking as a vendor perspective and check the figures against the official pricing pages linked in each provider section.

Criteria. Each provider is compared on the same attributes: archive, debug and trace access, published SOC 2 Type II and ISO 27001 certifications, dedicated infrastructure, pricing model and free tier, and support or SLA commitments. The order is editorial judgment for custody and asset-management workloads. It is not a benchmark result. The latency benchmark below covers only 3 of the 7 listed providers (Chainstack, Alchemy and Quicknode), so it informs the latency discussion for those three and does not determine the order.

Sources. Prices, limits and certifications come from each provider’s public pricing, documentation and security pages, as summarized on this page as of August 2026 unless a section says otherwise. Latency figures for Chainstack, Alchemy and Quicknode come from the Chainstack Labs Research Team’s October 2026 Ethereum benchmark. Chainstack’s platform uptime is shown as 99.99%+ platform uptime (status.chainstack.com, Ethereum Mainnet, 99.99%+ over the 90 days to October 29, 2026). That figure is platform uptime from the status page, which covers the Mainnet component of each chain (testnet and devnet components are tracked separately) and Chainstack only; it is separate from the benchmark’s probe success rate and from contractual SLAs, and it is not a guarantee.

Not covered. This article leaves out Infura, GetBlock and dRPC (general-purpose providers without both certifications published, as described in the comparison section; GetBlock publishes SOC 2 Type II only), and other Ethereum RPC providers may also exist. dRPC appears in the benchmark table because the report measured it, but it is not one of the 7 listed providers. RouteMesh, Uniblock, Blockdaemon and Ankr were not measured by the benchmark.

Latency data. See the Ethereum latency benchmark section for the figures, the method and what the benchmark does not show. For your own workload, measure from the region where your application runs; see how to measure RPC latency (P50, P95, P99).

💡 Already using Chainstack? Jump straight to the Ethereum tooling docs or deploy your endpoint in minutes at chainstack.com/build-better-with-ethereum.

Custodians and asset managers on Ethereum: RPC requirements

Custody infrastructure is not latency-critical the way trading bots are — a settlement confirmation taking 200ms instead of 50ms rarely matters. What matters is correctness, completeness, and auditability at scale. A missed eth_getLogs event is a reconciliation gap. A failed debug_traceTransaction call is an incomplete audit record. A provider outage during a regulatory reporting window is a compliance event, not a technical inconvenience.

Latency and throughput profile

Custody platforms typically run two concurrent workload patterns:

High-frequency polling — monitoring hundreds or thousands of wallet addresses for incoming transfers using eth_getLogs on short block ranges every few seconds. This is sustained, moderate-RPS load (roughly 50–200 RPS) that runs continuously.
On-demand audit calls — triggered by compliance requests, fund NAV calculations, or incident response. These are bursty, latency-tolerant calls that must succeed on the first attempt. debug_traceTransaction against an archive node is the canonical example.

The combination demands a provider with consistent throughput under sustained load and archive plus debug access that does not degrade under concurrent requests. A shared endpoint with aggressive method-level rate limits fails the second pattern at exactly the wrong moment — during an audit request.

Key RPC methods for custody and asset management

eth_getLogs — the reconciliation workhorse. Custody platforms use it to index ERC-20 Transfer events across client wallets and rebuild an audit log of every inbound and outbound flow. Ethereum’s recommended block range per call is capped by most providers; full historical reconciliation means calling it iteratively across the whole range, which requires a provider with no aggressive block-range caps and reliable response times at volume. eth_getLogs does not require an archive node for recent blocks — a full node serves event log scans on the last ~128 blocks of state, and log queries further back still work without full historical state access.
eth_getTransactionReceipt — settlement confirmation. This determines whether a transaction actually succeeded (status: 1) and, critically, what events it emitted via the logs array. Custody workflows call this on every outbound transaction and every incoming transfer flagged by eth_getLogs. It must return reliably with complete log data — truncation breaks downstream reconciliation.
eth_getTransactionByHash — audit record retrieval. Regulators and internal auditors routinely request full transaction detail: sender, receiver, value, input data, gas parameters. For transactions older than a full node’s retention window, this requires archive access.
debug_traceTransaction — the compliance trace. This returns the complete EVM execution trace of a transaction: every opcode, every storage read and write, every internal call. It’s mandatory for fund-flow analysis (multi-hop DeFi interactions), incident investigation, and regulatory audit requests. This method requires an archive node and is frequently disabled or rate-limited on shared endpoints because of its computational cost. For custody platforms building on RPC infrastructure for tokenized asset protocols, this is the method regulators tend to ask for first.
trace_transaction — the Erigon-namespace alternative. It returns structured internal-transaction traces (call traces, value transfers) without the opcode-level detail of debug_traceTransaction, which is often easier to parse for fund-flow analysis. Also requires archive access.
eth_getBalance — balance verification. Called on segregated client wallets for daily NAV calculations and proof-of-reserves attestations. Historical balance queries at a past block — for point-in-time NAV or audit reconstruction — require an archive node.
eth_blockNumber — chain-tip synchronization. Used to confirm the provider’s node is in sync before running reconciliation; a stale block height means stale data, a critical failure for settlement workflows.
eth_subscribe (logs) — WebSocket event streaming. Production custody platforms subscribe rather than poll for real-time settlement detection, filtering on specific contract addresses (stablecoin contracts, custody registries) for sub-block-latency delivery. Reconnect handling matters here — a dropped subscription means missed events.

For the full execution and consensus-layer method surface, including the debug_*, trace_*, and erigon_* namespaces referenced above, see the Ethereum API reference.

Infrastructure requirements

Archive node access is non-negotiable for any custody platform that serves historical data requests, responds to regulatory inquiries, or runs historical NAV calculations. debug_traceTransaction, trace_transaction, and historical state queries (eth_getBalance at a past block) all require it — see Chainstack’s archive data page for what full historical-state access looks like in practice. A full node keeps roughly the latest 128 blocks of state; anything older needs archive. Providers that offer archive as a capped or paid add-on with limited historical depth need to be evaluated on that depth specifically, not just “archive: yes/no.” One nuance worth knowing: pre-built native tracers (callTracer, prestateTracer) work against a standard archive node, but custom JavaScript tracers require a Dedicated Node rather than a shared archive endpoint.

Dedicated Nodes versus shared endpoints becomes relevant above roughly 100 RPS sustained, or whenever concurrent debug_traceTransaction calls need to run without competing for a shared rate-limit budget. Shared endpoints with method-level throttling can rate-limit trace calls under load — at precisely the moment a compliance team needs them.

MEV protection is a real, if underappreciated, custody concern on Ethereum — large custodial rebalancing transactions and client withdrawals sent through a public mempool are visible to searchers before they land, creating front-running and sandwich-attack exposure. A provider that routes transactions through a private order flow by default, rather than as an opt-in step a compliance-constrained team has to remember to configure, meaningfully reduces that exposure.

Geographic redundancy and SLA matter for regulated entities. A 99.9% SLA (roughly 8.7 hours of downtime per year) is thin for infrastructure supporting live fund operations. In our assessment, providers publishing a 99.99%+ SLA with contractual remedies are the more defensible choice for institutional deployment.

SOC 2 Type II and ISO 27001 certification are now the baseline for institutional vendor onboarding — banks and fund administrators require them as standard procurement criteria. A self-reported “SOC 2 in progress” claim is not equivalent to a completed Type II audit with a published report, and the two certifications cover different things: SOC 2 Type II evidences operating effectiveness over a review period, ISO 27001 evidences a certified information security management system. A provider holding one is not automatically holding the other, so check both explicitly rather than treating them as a single checkbox.

Ethereum RPC latency benchmark (October 2026): Chainstack, Alchemy and Quicknode

Custody workloads are less latency-sensitive than trading, but consistency matters: a P99 spike during end-of-day reconciliation is more damaging than a slightly higher average. This section shows the one dated, published measurement available for Ethereum, with its limits.

Disclosure: Chainstack operates the monitoring service, the public Grafana dashboard and the scoring formula behind this benchmark, and is one of the four providers measured. The figures below come from the Ethereum RPC provider benchmark (Research Note No. 050, published October 8, 2026) by the Chainstack Labs Research Team. Live data is on Chainstack Compare and the Chainstack performance dashboard. Compare labels the success-rate column “availability”; here it is called probe success rate because it is not platform uptime.

Method in brief: a 7-day window ending October 8, 2026; probes every 1 to 3 minutes from three cloud regions (Germany, US and Singapore) using a Python service on Vercel serverless functions; each provider on its first paid plan; a request counts as failed if it takes longer than 55 seconds. The score is response time divided by success rate cubed per region (lower is better), averaged equally across the three regions. The latency term is the mean of each method’s P95 and excludes eth_subscribe. The collection code is open source (compare-dashboard-functions).

Provider	Listed in this article	Score (lower is better)	Probe success rate	P50 (ms)	P95 (ms)	P99 (ms)	Germany P95 (ms)	US P95 (ms)	Singapore P95 (ms)
Chainstack	Yes	0.079	99.98%	36	78	117	47	143	42
Quicknode	Yes	0.097	99.98%	40	92	142	40	98	138
Alchemy	Yes	0.114	99.98%	38	103	188	64	80	164
dRPC	No (measured, not listed here)	0.670	99.96%	56	641	1,950	153	875	895

Source: Chainstack Labs Research Team, Ethereum RPC provider benchmark, October 2026 (7-day window ending October 8, 2026).

What the benchmark shows:

Among the three listed providers it measured, Chainstack has the lowest P50 (36 ms), P95 (78 ms) and P99 (117 ms) and the best overall score (0.079). The report puts second-place Quicknode’s score 23% higher.
Chainstack is not fastest everywhere. In the US, Alchemy (80 ms P95) and Quicknode (98 ms) beat Chainstack (143 ms). In Germany, Quicknode (40 ms) beats Chainstack (47 ms). Chainstack is lowest in Singapore (42 ms), and its best-to-worst region ratio is 3.4x (42 ms in Singapore against 143 ms in the US). A US-hosted custody or reconciliation system should weigh the US column, not the overall figure.
Chainstack, Quicknode and Alchemy tie on probe success rate at 99.98% (dRPC is 99.96%). This is the share of benchmark probe requests that succeeded over 7 days; it is not platform uptime and not an SLA.
dRPC, which is not one of the 7 listed providers, is degraded in the US (875 ms P95) and Singapore (895 ms P95).

What this benchmark does not show: RouteMesh, Uniblock, Blockdaemon and Ankr were not measured, so this page does not claim Chainstack is the fastest of the 7 listed providers. The report covers a single 7-day window, each provider on its first paid plan only (free, higher and dedicated plans can behave differently), and synthetic probes from three cloud regions rather than end-user latency. It excludes eth_subscribe, so it says nothing about WebSocket subscription latency or stability, and this article cites no per-method figures, so it is not evidence about debug_traceTransaction or archive-call performance specifically. It does not measure throughput under load, cost or correctness, and because Chainstack chose the scoring formula, a different formula could reorder close pairs.

Ethereum RPC rate limits and free tiers compared

Rate limits and billing units differ between providers, so the same number of calls can cost different amounts. This table uses only the figures given in the provider sections below; “Not stated here” means this article gives no figure.

Provider	Free tier	Free-tier rate limit	Paid tiers or flat-rate throughput	Billing unit
Chainstack	Permanent Developer tier, 3M RU/month (full node only, no card required)	25 RPS	Growth $49/month (20M RU, 250 RPS, archive and debug/trace); Pro $199/month (80M RU, 400 RPS, Dedicated Nodes); Business $499/month (200M RU, 600 RPS); custom Enterprise; Unlimited Node add-on from $149/month for 25 RPS	Request Units: 1 RU per full-node call, 2 RU for archive, debug and trace
RouteMesh	Not stated here	Not stated here	Per-request pricing, varying by key type, chain and method	Per request
Uniblock	Not stated here	Not stated here	Per-request pricing across 55+ providers	Per request
Blockdaemon	Not self-serve	Not applicable	Custom, institutional pricing	Quote-based
Ankr	Freemium, 200M API credits/month	Not stated here	Premium tier adds archive, WebSocket and dedicated private endpoints; prices not stated here	API credits
Alchemy	Permanent, 30M CU/month	15 RPS	Pay-as-you-go $0.52 per million CU (the docs show $0.525), no volume tiers; no standard dedicated tier	Compute units (method-weighted)
Quicknode	One-month trial, 10M credits	15 RPS	Build $49/month (80M credits, 50 RPS); Accelerate $249/month (450M credits, 125 RPS); Scale $499/month (950M credits, 250 RPS); Dedicated infrastructure on Enterprise	Credits (method-weighted)

Chainstack alternatives for custody and asset management on Ethereum

If Chainstack is not the right fit, these are the head-to-head differences between Chainstack and each alternative, using the facts in the sections of this article:

Chainstack vs RouteMesh: RouteMesh adds automatic failover across upstream providers and covers 1,000+ EVM networks, but archive and debug/trace depth depend on the upstream that answers and it publishes no SOC 2 or ISO 27001 report of its own; Chainstack runs its own archive and debug/trace stack and holds both certifications.
Chainstack vs Uniblock: Uniblock offers Unified APIs for token, NFT and market data and pooled rate-limit headroom across 55+ providers, but publishes no certification of its own and archive availability depends on the backend that serves a call; Chainstack offers full archive and debug/trace on a single operator with Dedicated Nodes.
Chainstack vs Blockdaemon: Blockdaemon holds both certifications and bundles custody, staking and RPC for large institutions, but has no self-serve signup or public pricing; Chainstack publishes its plans and can be deployed from the console in minutes.
Chainstack vs Ankr: Ankr has a larger free credit allowance (200M API credits/month on Freemium) and SOC 2 Type II, but ISO 27001 is not yet completed and it has no named enterprise support tier or public contractual SLA described here; Chainstack holds both certifications.
Chainstack vs Alchemy: Alchemy has a larger free tier (30M CU/month), NFT and Token APIs, SOC 2 Type II and a faster US P95 (80 ms against 143 ms) in the October 2026 benchmark, but publishes no ISO 27001, has no standard dedicated tier and weights archive calls by method; Chainstack bills a flat 2 RU for archive, debug and trace and had the lower overall P50, P95 and P99 in that benchmark.
Chainstack vs Quicknode: Quicknode holds both certifications, includes archive on all paid plans and had faster Germany (40 ms against 47 ms) and US (98 ms against 143 ms) P95 in the October 2026 benchmark, but its dedicated infrastructure is Enterprise-only, its free offering is a one-month trial and it uses method-weighted credits; Chainstack offers Dedicated Nodes from the Pro plan and had the lower overall P50, P95 and P99.

In-depth reviews: 7 Ethereum RPC providers for custodians and asset managers (2026)

1. Chainstack: best for production custody that needs compliance and cost predictability

Verdict: Full archive with debug/trace from the Growth plan, flat 1–2 RU pricing, SOC 2 Type II and ISO 27001, MEV protection on by default and a contractual SLA on Enterprise. Trade-off: Unlimited Node RPS tiers can cost more than per-request billing for spiky traffic, very large archive backfills need a sizing conversation, and its US P95 (143 ms) trailed Alchemy and Quicknode in its own October 2026 benchmark. Official pricing.

Custody workloads live or die on archive depth and cost predictability under audit-heavy trace usage. Chainstack maps onto both directly: every plan from Growth up includes full Ethereum archive access, with debug_traceTransaction, trace_transaction, and the full erigon_* namespace available on archive-enabled endpoints without extra method-level throttling beyond the plan’s RPS ceiling.

In our assessment, the pricing model is among the most transparent in this comparison (Chainstack publishes this page): a full node call costs 1 request unit (RU), and any call touching state 128+ blocks behind the tip — or any debug_*/trace_* call regardless of block age — costs 2 RU, flat. A debug_traceTransaction call costs the same as any other archive read in RU terms, which is what makes cost forecasting realistic for a compliance team budgeting an audit-heavy quarter. Plans run from the free Developer tier (3M RU/month, 25 RPS, full node only, no card required) through Growth ($49/month, 20M RU, 250 RPS, archive and debug/trace unlocked), Pro ($199/month, 80M RU, 400 RPS, Dedicated Nodes and SSO unlocked), and Business ($499/month, 200M RU, 600 RPS, role-based access control), with custom Enterprise contracts for larger volumes and contractual SLA terms.

For regulated workflows, Chainstack holds both SOC 2 Type II (completed December 22, 2025) and ISO 27001 (achieved July 2026) — the dual-certification pairing that vendor risk teams increasingly treat as table stakes. Global Nodes run geo-load-balanced across Chainstack’s Asia-Pacific, Europe, and US regions with 99.99%+ platform uptime on the public status page (not a contractual SLA), and MEV protection is switched on by default at node deployment for Ethereum mainnet — not an opt-in step — free on every plan including the Developer tier, and can be turned off from the node’s Add-ons tab for workflows that specifically need public-mempool behavior. The Enterprise plan adds SSO, granular role-based permissions, a dedicated account team, and a contractual uptime SLA (99.9% quarterly) with service credits — the package vendor risk reviews expect to see. Support scales with plan tier too: Developer and Growth get community and standard email support, while the Professional and Premium support add-ons commit to under-6-hour and under-1-hour response times respectively, with the Enterprise plan’s dedicated account team providing 24/7 coverage.

Dedicated Nodes isolate a custody workload from shared infrastructure — useful for platforms running continuous high-volume eth_getLogs polling alongside on-demand trace calls that would otherwise compete for the same rate-limit budget. For predictable-throughput monitoring pipelines, the Unlimited Node add-on (from $149/month for a 25 RPS tier, available from Growth up) bills a flat monthly rate instead of per-request, which removes the RU-forecasting exercise entirely for that workload.

The following snippet reconciles ERC-20 Transfer events into a monitored set of custody wallets over the last 500 blocks, using ethers.js in the style shown in the Ethereum tooling documentation:

import { ethers } from "ethers";

const provider = new ethers.JsonRpcProvider("YOUR_CHAINSTACK_ENDPOINT");

// ERC-20 Transfer event topic
const TRANSFER_TOPIC = ethers.id("Transfer(address,address,uint256)");

// Custody wallet addresses (padded to 32 bytes for topic filter)
const wallets = [
  "YOUR_CUSTODY_WALLET_1",
  "YOUR_CUSTODY_WALLET_2",
].map((addr) => ethers.zeroPadValue(addr, 32));

async function reconcile() {
  const latestBlock = await provider.getBlockNumber();
  const fromBlock = latestBlock - 500;

  const logs = await provider.getLogs({
    fromBlock,
    toBlock: latestBlock,
    topics: [TRANSFER_TOPIC, null, wallets],
  });

  for (const log of logs) {
    const amount = ethers.toBigInt(log.data);
    console.log(
      `Inbound transfer: ${log.transactionHash} | block ${log.blockNumber} | amount ${ethers.formatUnits(amount, 18)}`
    );
  }
}

reconcile();

Limitations: The Unlimited Node add-on’s flat RPS tiers can cost more than per-request billing for workloads with genuinely variable traffic that only occasionally spikes above the chosen tier. Very large one-off archive backfills are worth a sizing conversation with the Chainstack team rather than assuming a shared endpoint handles them unthrottled. In Chainstack’s own October 2026 Ethereum benchmark, Chainstack had the lowest overall P50, P95 and P99 of the three listed providers measured, but its US P95 (143 ms) was slower than Alchemy (80 ms) and Quicknode (98 ms), which matters for US-hosted monitoring and reconciliation systems.

Fit for custodians and asset managers: Excellent — full archive and debug/trace from the Growth plan, both certifications published, flat-rate cost model, MEV protection on by default, and an Enterprise tier built around SSO, RBAC, and contractual SLA.

2. RouteMesh: best as failover on top of a certified primary provider

Verdict: Routing layer with automatic failover across upstream providers. Trade-off: no SOC 2 or ISO 27001 report of its own, and archive and debug/trace depth depend on the upstream that answers. Pricing: no official pricing link has been verified for this page.

RouteMesh is an intelligent RPC routing layer spanning 1,000+ EVM networks, including Ethereum mainnet and Sepolia. It aggregates and routes across multiple underlying providers with sub-10ms routing decisions, real-time health scoring, and automatic failover — a different product shape from a direct node connection, closer to a reliability layer sitting in front of several backends at once.

Reference customers include LI.FI (60+ chains, 97% of its EVM RPC calls routed through RouteMesh), Vaults.fyi, and DAMM Capital — the latter a relevant reference point for asset-management buyers evaluating failover behavior under real fund-operations load. Pricing is per-request, varying by key type, chain, and method; an MCP server and Cursor integration are available for teams wiring routing decisions into agentic workflows.

Limitations: RouteMesh does not itself publish a SOC 2 or ISO 27001 report — its own security posture is a separate due-diligence item from whichever backend node operator ultimately serves a given request. Archive and debug/trace depth depend entirely on which upstream provider a request lands on, which makes it harder to give a single, consistent audit-workflow answer.

Fit for custodians and asset managers: Good for the specific job of failover — a custody platform running continuous wallet-monitoring benefits from automatic rerouting around a degraded backend, though it is a complement to a certified primary provider rather than a substitute for one.

3. Uniblock: best for enriched token and portfolio APIs

Verdict: Unified APIs plus pooled rate-limit headroom across 55+ providers. Trade-off: no independently published certification and archive/debug availability depends on the backend that serves each call. Official pricing.

Uniblock is a managed multi-chain infrastructure layer covering 300+ blockchains, including Ethereum, through 55+ underlying RPC providers with automatic routing, failover, and hedging. Beyond raw JSON-RPC, it exposes Unified APIs for token, NFT, and market data — useful for a custody or asset-management dashboard that needs enriched position data alongside settlement confirmation, without hand-rolling multiple integrations.

Customers include Plume Network, Stellar, Hypernative, and Oku Trade, across a reported 4,000+ developers and 3,000+ projects; the company raised $5.2M in March 2026. An MCP server and agent skills for Cursor and GitHub Copilot are available for teams building agentic tooling on top of the API layer.

Limitations: As with RouteMesh, Uniblock does not independently publish SOC 2 or ISO 27001 certification, and archive/debug availability is a function of whichever of its 55+ backend providers serves a given call — not a fixed, auditable guarantee a compliance team can cite on its own.

Fit for custodians and asset managers: Good where the enriched Unified APIs genuinely save integration work — a portfolio-reporting layer built on top of raw settlement data is a real fit; treat the underlying RPC reliability the same way you’d treat any aggregator, as a complement rather than the certified system of record.

4. Blockdaemon: best for large institutions with a formal vendor-onboarding process

Verdict: Custody, staking and RPC under one contract, with SOC 2 Type II and ISO 27001. Trade-off: no self-serve signup or public pricing, so evaluation starts with a sales conversation. Official site (pricing on request).

Blockdaemon sells directly to the exact buyer this article is written for: it serves 400+ institutions, including major banks and payment networks, and bundles production-grade Ethereum RPC endpoints with custody, staking, and tokenization infrastructure rather than offering RPC as a standalone self-serve product. High-availability node clusters with automated scaling and failover back the RPC layer.

Both SOC 2 Type II and ISO 27001 certifications are published, matching Chainstack and Quicknode as the only providers in this comparison holding both. Pricing is custom and institutional rather than self-serve — there’s no public plan tier to point to, which is itself consistent with the buyer Blockdaemon targets: procurement teams that expect a commercial negotiation, not a credit card checkout.

Limitations: No self-serve signup or public pricing means evaluation requires a sales conversation before a team can even see representative costs — a meaningfully slower path than deploying an endpoint from a console in minutes. Not a fit for teams that want to prototype before committing to a vendor relationship.

Fit for custodians and asset managers: Excellent for large institutions that already run a formal vendor-onboarding process and want custody, staking, and RPC bundled under one contract; less practical for a team that wants to deploy and test an endpoint today.

5. Ankr: best for evaluation and pre-production on a large free allowance

Verdict: SOC 2 Type II and a Freemium plan of 200M API credits/month. Trade-off: ISO 27001 is not yet completed, the distributed operator model may vary response times and no public contractual SLA is described here. Pricing: no official pricing link has been verified for this page.

Ankr holds SOC 2 Type II certification and runs a distributed node-operator model across 80+ chains, with the largest free credit allowance in this comparison: a Freemium plan offering 200M API credits per month (credit units differ between providers, so the figures are not directly comparable). Archive access, WebSocket support, and dedicated private endpoints are available from the Premium tier up.

ISO 27001 certification is not yet published as completed — as of this writing it’s described as a program underway rather than an achieved certification, which puts Ankr a step behind the dual-certified providers in this comparison on the compliance axis. The distributed operator model can also introduce more response-time variability than a single-operator infrastructure stack, a bigger concern for correctness-sensitive custody reconciliation than it would be for a latency-tolerant workload.

Limitations: ISO 27001 not yet completed. Distributed node-operator model may introduce response-consistency variability. No named enterprise support tier or public contractual SLA comparable to Chainstack’s or Quicknode’s.

Fit for custodians and asset managers: Moderate — the free tier is genuinely useful for evaluation and pre-production work, but the incomplete compliance pairing and infrastructure-consistency profile make it a harder sell for a production custody deployment.

6. Alchemy: best for developer tooling and a permanent free tier

Verdict: 30M CU/month free tier, NFT and Token APIs and SOC 2 Type II; fastest US P95 (80 ms) of the three providers measured in the October 2026 benchmark. Trade-off: no published ISO 27001, no standard dedicated tier and method-weighted CU billing for archive-heavy work. Official pricing.

Alchemy is widely used among Ethereum developers, with archive access and debug/trace support on paid plans. Its platform layer — NFT APIs, Token APIs, and enriched transaction data — adds real value for custody platforms building portfolio views, since it abstracts some multi-call reconciliation patterns into single API calls rather than requiring them to be hand-assembled from raw logs.

Pay-as-you-go pricing runs $0.52 per million compute units (CU) with no volume tiers (the docs show $0.525). The free tier (30M CU/month, 15 RPS) is a permanent plan rather than a trial — a point in Alchemy’s favor for evaluation and pre-production use. Alchemy holds SOC 2 Type II certification but does not publish ISO 27001; for custody teams whose vendor risk policy requires both, that’s a specific, nameable gap rather than a general compliance shortfall. There is no standard self-serve dedicated-node tier — isolated infrastructure requires an enterprise agreement with custom pricing.

Budget-conscious teams scaling past the free tier should model Alchemy’s CU costs against their actual trace and archive call volume before committing, since method-weighted billing means the effective cost per audit-heavy call runs above the headline per-CU rate.

Limitations: No published ISO 27001 is a specific, nameable gap for dual-certification procurement requirements. No standard dedicated-node tier. Method-weighted CU billing makes archive-heavy audit workloads costlier than the sticker price implies.

Fit for custodians and asset managers: Moderate — strong developer tooling and a permanent free tier, held back for regulated custody procurement by the missing ISO 27001 certification and the absence of a standard dedicated tier.

7. Quicknode: best for dual-certified infrastructure with self-serve signup

Verdict: SOC 2 Type II plus ISO 27001, archive on every paid plan; fastest Germany P95 (40 ms) of the three providers measured in the October 2026 benchmark. Trade-off: dedicated infrastructure and SSO are Enterprise-only, credits are method-weighted and the free offering is a one-month trial. Official pricing.

Quicknode includes archive access on all paid Ethereum plans, and debug_traceTransaction and WebSocket subscriptions are available across the paid tier ladder. It holds both SOC 2 Type II and ISO 27001 certification, matching Chainstack and Blockdaemon as the only providers here with the full dual-certification pairing.

The free option is a one-month trial (10M credits, 15 RPS) rather than a permanent standalone plan — a real, citable distinction from Chainstack’s and Alchemy’s permanent free tiers. Paid plans run from Build ($49/month, 80M credits, 50 RPS) through Accelerate ($249/month, 450M credits, 125 RPS) and Scale ($499/month, 950M credits, 250 RPS) up to several Business+ tiers, with Dedicated infrastructure and SSO reserved for Enterprise agreements. Pricing uses method-weighted credits, where trace and archive calls consume more credits than a standard read — teams running heavy audit-trace volume should model that against their actual call mix rather than the headline plan price, since it’s the same structural pattern that makes Alchemy’s sticker price understate archive-heavy costs.

Limitations: Method-weighted billing makes cost forecasting harder for audit-heavy trace workloads than Chainstack’s flat RU model. Dedicated infrastructure and SSO are Enterprise-only, so mid-market custody platforms share endpoints with the broader user base unless they negotiate an Enterprise contract.

Fit for custodians and asset managers: Strong — archive on every paid plan and the same dual-certification pairing as Chainstack and Blockdaemon, offset by a time-limited free tier and a cost model that climbs faster than the headline price suggests once trace-heavy workloads scale.

Getting started with Ethereum on Chainstack

Deploy a production Ethereum endpoint in a few minutes and connect it to a custody or reconciliation pipeline:

Log in to the Chainstack console (or create an account).
Create a new project
Select Ethereum as your blockchain protocol
Choose network: Ethereum Mainnet, Sepolia, or Hoodi testnet
Deploy the node
Open Access and credentials and copy your HTTPS and WebSocket endpoints

For custody workloads running continuous eth_getLogs polling alongside on-demand trace calls, evaluate Dedicated Nodes after the initial deploy — both live inside the same project, so moving from a shared Global Node to isolated infrastructure doesn’t require re-architecting the integration.

The Ethereum tooling documentation covers setup for MetaMask, Remix, viem, Web3.py, ethers.js, Hardhat, and Foundry. See the Chainstack provider entry above for a worked eth_getLogs reconciliation example.

Need testnet ETH for integration testing? Grab some from the Chainstack Sepolia faucet or the Chainstack Hoodi faucet — both distribute up to 0.5 ETH every 24 hours. Note that Hoodi is archive-only on Chainstack; Sepolia supports both full and archive nodes.

🤖 You can also access Chainstack Ethereum RPC directly from Claude, Cursor, Codex, Windsurf, Gemini CLI, GitHub Copilot, Antigravity, Claude.ai, or ChatGPT using Chainstack MCP. For a fuller agent stack — MCP, the Chainstack skill, llms.txt for context ingestion, and WebMCP for agentic browsers — see the Chainstack Agents page.

⚡ Performance tip: For custody platforms running continuous eth_getLogs polling across hundreds of addresses, the Unlimited Node add-on bills a flat monthly rate instead of per-request. At 25–100 RPS sustained, the fixed cost can be lower than per-request billing with overage; compare both against your actual call volume.

Conclusion

For Ethereum custody and asset management in 2026, in our assessment the decision comes down to one question: which providers have a published, current, dual-certification story (SOC 2 Type II and ISO 27001) alongside flat-rate archive access that doesn’t punish audit-heavy months.

For production custody deployments prioritizing compliance and cost predictability together: in our assessment, Chainstack (published by Chainstack, see how we compared) — full archive and debug/trace from the Growth plan, both certifications published, flat 1–2 RU pricing regardless of method, and MEV protection on by default
For large institutions running a formal vendor-onboarding process: Blockdaemon — custody, staking, and RPC bundled under one contract with both certifications published, at the cost of a slower, sales-led evaluation path
For dual-certified infrastructure with a self-serve signup: Quicknode — the same certification pairing as Chainstack, with a higher effective cost at scale for trace-heavy audit workloads
For evaluation, prototyping, and pre-production environments: Alchemy or Ankr — solid technical capability and generous free tiers; each carries one specific, nameable compliance gap (missing ISO 27001) worth flagging before a production commitment
For resilience layered on top of a certified primary provider: RouteMesh or Uniblock — useful for multi-provider failover and enriched portfolio APIs, but not a substitute for a directly auditable node operator

Deploy your Ethereum RPC endpoint on Chainstack →

Frequently asked questions

Q: Does debug_traceTransaction require an archive node on Ethereum?

Yes. debug_traceTransaction and trace_transaction both require an archive node because they reconstruct the full EVM state at the time of execution. A full node keeps roughly the latest 128 blocks and prunes state beyond that, so it cannot serve either method for older transactions. Confirm archive and debug/trace access explicitly before signing a provider contract — don’t assume it’s included just because the provider lists “archive” as a feature.

Q: Which Ethereum RPC providers offer a permanent free tier for custody platform development and testing?

Chainstack (3M RU/month, no card required), Alchemy (30M CU/month), and Ankr (200M credits/month on its Freemium plan) all run permanent free tiers. Quicknode’s free option is explicitly a one-month trial rather than a standing plan. For a custody platform evaluating vendors before committing budget, a permanent free tier removes the pressure to make a compliance decision on an artificial deadline.

Q: How do I migrate a custody platform from a general-purpose RPC provider to a certified one without downtime?

Run both endpoints in parallel behind a feature flag or load balancer, mirror eth_getLogs and eth_getTransactionReceipt traffic to the new provider first to validate reconciliation output matches exactly, then cut over write traffic (transaction submission) last, once read parity is confirmed over a full settlement cycle. Verify the new provider’s archive depth actually covers your existing audit-trail retention requirement before decommissioning the old endpoint — don’t assume equivalent depth.

Q: How do I benchmark Ethereum RPC provider latency and consistency for custody workloads?

Custody workloads care more about consistency than average latency — a p99 spike on eth_getLogs during end-of-day reconciliation is more damaging than a slightly higher mean on eth_blockNumber. Test with your actual method mix: eth_getLogs at your typical block range, debug_traceTransaction on representative transactions, and WebSocket subscription stability over 24 hours. In Chainstack’s October 2026 Ethereum benchmark (a 7-day, Chainstack-run test of Chainstack, Alchemy, Quicknode and dRPC), the overall P99 was 117 ms for Chainstack, 142 ms for Quicknode and 188 ms for Alchemy, with a 99.98% probe success rate for all three; the benchmark does not cover WebSocket subscriptions or trace methods. compare.chainstack.com tracks live cross-provider Ethereum performance for a quick top-level check; the Chainstack performance dashboard gives the same data broken down by method and region.

Q: Does MiCA compliance require my Ethereum RPC provider to be SOC 2 certified?

MiCA doesn’t name SOC 2 specifically — it requires custodians to maintain “robust operational and technical” controls and to run third-party vendor risk assessments. In practice, institutional counterparties (prime brokers, fund administrators, bank custody partners) increasingly require SOC 2 Type II and ISO 27001 as part of vendor onboarding regardless of what the regulation itself names. Choosing an uncertified provider doesn’t violate MiCA directly, but it lengthens the vendor risk assessment and may require compensating-controls documentation instead.

Q: What’s the real cost difference between Chainstack’s flat RU pricing and a method-weighted provider for a trace-heavy custody workload?

Chainstack bills 1 RU per full-node call and 2 RU per archive/debug/trace call, regardless of method — a flat multiplier with no per-method differentiation. Method-weighted providers charge substantially more for heavy calls: Per the vendors’ own pages, Alchemy lists debug_* at 40 CU and eth_getLogs at 60 CU against 10 CU for eth_blockNumber, and Quicknode applies a 2x multiplier to trace and debug calls (4x to specialized calls) on its 20-credit Ethereum base; see also Chainstack’s own cost comparison. For a workload with a meaningful share of debug_traceTransaction or historical-state calls, that gap compounds fast — model your actual method mix against each pricing page rather than comparing headline plan prices.

Additional resources

Changelog

We review this comparison regularly so that the information stays accurate, clear, and easy to verify. This page was reviewed to improve how the comparison is structured and sourced.

October 29, 2026: Added a quick verdict, a Rank and Best for column in the provider comparison table (moved up), a methodology and disclosure section, an Ethereum latency benchmark section (October 2026, covering Chainstack, Alchemy and Quicknode), a rate limits and free tiers comparison, alternatives guidance, numbered provider sections with one-line verdicts and official pricing links, corrected vendor pricing, limits and certification details, corrected vendor pricing, limits and certification details, and wording clarifications.
August 17, 2026: Original publication.

SHARE THIS ARTICLE