Top 9 Plasma RPC Providers 2026: Pricing, Archive and Log Limits Compared

A USDT0 transfer on Plasma uses 60,924 gas. At the measured base fee of 8 wei plus a 1 gwei tip, that is about 0.0000609 XPL, or about $0.000006 at $0.101 per XPL. Gas on Plasma is close to free.

The RPC calls around that transfer are not. A payment flow makes 5-8 calls: a balance read, a nonce lookup, a gas estimate, the send, a receipt poll or subscription, and a log confirmation. At Alchemy's $0.525 per 1M compute units and a weighted average of about 25.9 CU per Plasma call, one call costs about $0.0000136, so 5-8 calls cost $0.00007-0.00011, roughly 11-18 times the gas.

At Dwellir's $2-5 per 1M responses, the same calls cost $0.00001-0.00004 per payment, about 2-7 times the gas. On Plasma, the RPC bill is the infrastructure cost that grows with payment volume.

This comparison covers 9 providers selling managed Plasma RPC, plus three smaller options. Pricing is a list-price snapshot from 2026-09-28, normalized to dollars per million requests, alongside archive, trace and eth_getLogs limits.

What Makes Plasma Different for RPC

Plasma here means the stablecoin payments Layer 1 at chain ID 9745, not the 2017 Ethereum scaling design of the same name. It is an EVM chain that pairs PlasmaBFT, a pipelined Rust implementation of Fast HotStuff, with a Reth execution layer. The public endpoint reported reth/v1.11.3 on 2026-09-28. Solidity, Foundry, Hardhat, viem, ethers.js and MetaMask work unchanged.

Mainnet beta launched on September 25, 2025, and the docs still call it "Plasma Mainnet Beta". Consensus runs on a small validator set chosen by Plasma; external validators and delegation are not live.

One-second blocks and a finalized tag two blocks back

Measured block time is 1 second, and the finalized tag trailed latest by 2 blocks. Plasma's site claims "3s deterministic finality". Confirm payments against finalized rather than a block-count heuristic. At 86,400 blocks per day, slow ingestion falls behind quickly.

Gas is XPL only, and almost all of it is the tip

Fees are paid in XPL (18 decimals) under EIP-1559, with the base fee burned; custom ERC-20 gas is announced but not live. The base fee sits at 8 wei because blocks are about 5% full (36M gas limit, about 1.7M used), so almost all of the cost is the 1 gwei tip. Wallets that hard-code Ethereum-style tips overpay but still pay fractions of a cent.

USDT0 makes eth_getLogs the method to compare

USDT0 is the main asset on Plasma: 0xB8CE59FC3717ada4C02eaDF9682A9e934F625ebb, 6 decimals, with 885.6M in supply measured on 2026-09-28. Bitfinex and Tether CEO Paolo Ardoino are among Plasma's investors.

USDT0 Transfer logs average about 2.2 per block, so a 10,000-block window (about 2.8 hours) already exceeds the public endpoint's 20,000-result cap. Backfills need 1,000-5,000 block windows or a provider with higher caps. See the eth_getLogs cost traps guide for how range caps and per-call weights interact.

Zero-fee USDT0 is a relayer, not an RPC feature

Zero-fee USDT0 transfers run through Plasma's own relayer: users sign an EIP-3009 transferWithAuthorization, and the Plasma Foundation pays the gas. At launch this was limited to Plasma's own products, and third-party access is unverified as of September 2026. No RPC provider can sell you zero-fee USDT0, and you still need normal RPC to read balances and authorizationState and to watch Transfer and AuthorizationUsed logs.

Account abstraction and trace support

ERC-4337 and EIP-7702 are supported, and handleOps calls appear in measured blocks. Following the internal transfers inside a handleOps bundle or a router call needs debug or trace methods, and support for those varies by provider. Custom gas tokens, confidential payments and dedicated payment lanes are on Plasma's roadmap and not live.

Ecosystem context

DeFiLlama showed Plasma TVL at $546M on 2026-09-28, against a peak of $6.36B on 2025-10-09, down about 91% after incentive programs ended. About $1.40B in stablecoins sits on chain, Aave V3 holds $469M, and measured activity is roughly 587,000 transactions per day. The Plasma One neobank and Visa card launched publicly around June 17-18, 2026.

What to Look for in a Plasma RPC Provider

Log query capacity. USDT0 reconciliation runs on eth_getLogs, and the public endpoint caps it at 10,000 blocks and 20,000 results. Check each provider's block-range cap, result cap and per-call charge.

Finality-aware reads. On Plasma, finalized should trail latest by about 2 blocks. Ask whether the provider serves finalized from a node in sync with consensus, because a lagging node reports stale finality.

WebSocket. newHeads fires every second, and a USDT0 logs subscription delivered 22 events in 10 seconds when measured. Confirm eth_subscribe works for both on your plan.

Archive access. Reconciliation and audits call eth_call and eth_getBalance at old blocks, which needs archive state. A Plasma node uses only about 350 GB of storage, so archive is cheap for providers to run; check whether it is on the shared plan or only on dedicated nodes.

Debug and trace. Tracing internal transfers in 4337 bundles and router calls needs debug_* or trace_*, often only on paid plans. Test debug_traceTransaction and trace_block on a handleOps transaction before you commit.

Pricing unit. Per-request billing and weighted units produce very different bills for the same traffic: eth_getLogs is 1 unit on a per-request plan and 60 CU on Alchemy. See RPC providers without compute units for the general case.

Uptime commitment. A 99.9% SLA allows 8.7 hours of downtime a year; 99.99% allows 52 minutes.

Compliance documents. If your counterparties are payment institutions, SOC 2 and ISO 27001 reports become part of vendor qualification.

How to Compare Plasma Pricing Across Unit Systems

Every headline rate in this comparison is quoted in a different unit. The only way to compare them is to convert each one to dollars per million calls for the methods a payment backend actually makes, using the weight each provider assigns to the method.

Per-request billing. One response is one unit, whatever the method. Dwellir, thirdweb and NOWNodes bill this way, and Chainstack does too on recent state (1 request unit per call, 2 for reads 127 or more blocks behind the tip and for every debug or trace call). Dwellir's Developer plan works out to $1.96 per 1M calls included, with overage at $2-5 per 1M. Chainstack Growth is $2.45 per 1M included, $15 per 1M overage.

Compute units and credits. The provider charges per million units and gives each method a weight, so the cost of a call depends on the method.

• Alchemy charges $0.525 per 1M CU. On Plasma that is $13.65 per 1M eth_call, $31.50 per 1M eth_getLogs, $21.00 per 1M eth_sendRawTransaction, $10.50 per 1M eth_getTransactionReceipt or eth_getBalance, and $21.00 per 1M debug_traceTransaction.
• QuickNode assigns a flat 20 credits to every Plasma call, which is $10.00 per 1M calls at Business's overage rate and $12.40 at Build's. Debug, trace and txpool calls carry a 2x chain multiplier, $20.00 per 1M at the Business overage rate, and large calls such as txpool_content are 4x, $40.00 per 1M.
• dRPC assigns 20 CU to every Plasma method, a flat $6.00 per 1M calls, trace and archive included.
• Tatum's Scale plan works out to $14.28 per 1M standard calls, $35.70 per 1M eth_call and $357 per 1M debug or trace calls.
• Tenderly weights eth_call at 4 TU, writes at 20 and debug or trace at 40, but publishes no dollar rate, so there is no per-call figure to compare.

Method by method, in dollars per 1M calls at each plan's included rate:

Dwellir and Chainstack overage runs $2-5 and $15 per 1M respectively, and Chainstack doubles the unit for reads 127 or more blocks behind the tip and for debug and trace calls.

The monthly totals later in this post use a mostly-reads payment mix: 40% eth_call, 25% receipts, 15% eth_blockNumber, 10% eth_getLogs over small ranges, 5% eth_getBalance and 5% eth_sendRawTransaction, all on recent blocks. On that mix a Plasma call averages 25.9 CU on Alchemy (eth_blockNumber is 10 CU), so 1M calls cost $13.60 and the free 30M CU covers about 1.2M calls. On QuickNode every call is 20 credits, so Build's 80M credits cover 4M calls and Business's 2B credits cover 100M calls at $999. Tatum averages 3.2 credits per call (5 for eth_call, 2 for the rest), or $22.85 per 1M at Scale's 350M-credit step.

1. Dwellir

Dwellir serves Plasma on 1:1 pricing: one RPC response is one API credit, with no compute units, including on debug methods. The endpoint is https://api-plasma-mainnet.n.dwellir.com/<your-api-key>, with WSS on the same host and eth_subscribe for newHeads working. See the Plasma network page for setup details.

Key features:

• WebSocket on the same endpoint
• debug_* and txpool_* exposed on the shared Plasma endpoint
• Unlimited Node for flat, unmetered billing on the same URL and key
• 140+ networks, European infrastructure, autoscaling
• Crypto payment on every paid plan

Pricing:

A dedicated Plasma node cluster lists at $3,736 per month and includes archive, redundant nodes, regions of your choice, a 99.99% SLA and no request counting.

Considerations: Dwellir runs Plasma as full nodes on the shared endpoints, not archive. Historical eth_call at old blocks is not guaranteed, and archive on Plasma is only available through a dedicated node cluster. Trace methods (trace_*) are not available on the shared Plasma endpoint; debug_* and txpool_* are. Chainstack, QuickNode, dRPC and OnFinality all advertise archive on Plasma.

On recent-state traffic, the flat unit is the advantage: eth_call, receipts and eth_getLogs cost 20-60 units each on Alchemy and QuickNode and one credit here. Scale's 5,000 RPS sustained matches dRPC Growth and is 10 times QuickNode Business at 500 RPS.

Best for: High-volume payment reads and sends on recent state where predictable per-response pricing matters. Teams that need archive should pair Dwellir with an archive provider or use a dedicated cluster.

2. Alchemy

Alchemy lists Plasma mainnet with archive data, WebSocket, and debug and trace APIs, and appears on Plasma's partners page. Endpoints are https://plasma-mainnet.g.alchemy.com/v2/<api-key> and the matching wss:// URL. There is no keyless public endpoint.

Key features:

• "Full archive data" listed on all plans
• WebSocket on Plasma
• Debug and Trace APIs on Pay as You Go (not on Free)
• Bundler API, Gas Manager and Webhooks live on Plasma
• Transfers, Token and Prices APIs marked "Request support" (not live on Plasma)
• 100+ blockchains; signed SLAs on Enterprise

Pricing:

Extra PAYG throughput costs $160 per month per 5,000 CU/s block, up to a 30,000 CU/s self-serve cap.

Alchemy's strength on Plasma is the live account abstraction stack for 4337 flows. The trade-off is weighting: eth_getLogs costs 60 CU, or $31.50 per 1M log queries, and the payment mix lands at $13.60 per 1M requests. The free 30M CU is about 1.2M Plasma requests, without debug or trace. Crypto payment is not stated on the pricing page.

Best for: Teams building ERC-4337 payment flows who want bundler, gas sponsorship and RPC from one vendor and can model their CU consumption.

3. QuickNode

QuickNode lists Plasma mainnet with HTTP and WSS, archive and no pruning, marked "Beta", and is first on Plasma's official RPC providers page. There is no keyless endpoint.

Key features:

• Archive on Plasma
• Debug and Trace APIs on paid plans
• Streams and Webhooks for Plasma
• SOC 1/2 Type 2 and ISO 27001
• Pay in crypto on paid plans
• 79+ chains

Pricing:

Effective price is the plan cost spread over the Plasma calls its credits cover.

Annual billing lowers each tier, for example Build to $34 per month.

At 20 credits per call, the credit totals on the pricing page overstate Plasma request capacity by 20 times, and debug, trace and txpool calls cost 40 to 80. At Business rates a Plasma request costs $10.00 per 1M, which buys archive, trace, Streams and a full compliance set. Support response SLAs run from 24 hours (Build) to 8 hours (Scale, Business), with a contractual uptime SLA on Enterprise and no permanent free tier.

Best for: Regulated teams that need archive, trace and SOC reports together and can budget 20 credits per call.

4. Chainstack

Chainstack deploys Plasma mainnet and testnet archive nodes with Debug and Trace APIs as Global Nodes. It bills request units with no method weights, which puts it closest to Dwellir on unit design.

Key features:

• Archive on Plasma (Global and Dedicated nodes)
• trace_block, trace_call, trace_replayBlockTransactions and debug_traceCall documented for Plasma
• WebSocket on all plans
• Unlimited Node add-on from $149/month at 25 RPS
• Self-hosted Plasma deployments (full mode) and dedicated nodes from Pro
• SOC 2 Type II and ISO 27001; 70+ chains; crypto payment

Pricing:

A request under 127 blocks behind the tip costs 1 RU; 127 or more blocks behind costs 2 RU as an archive read, and debug and trace calls cost 2 RU regardless of block age. Debug, trace and archive data start on Growth.

Chainstack combines archive, trace, SOC 2 and 1:1 recent-block pricing on one shared endpoint. The costs sit in overage ($10-20 per 1M below Enterprise) and in old-state reads at 2 RU. Its Unlimited Node is priced above Dwellir's at each published tier ($149 vs $119 at 25 RPS, $649 vs $479 at 100 RPS), and the page does not make clear whether the add-on sits on top of the Growth subscription. The Plasma page claims 99.99%+ uptime, with a formal uptime SLA on Enterprise.

Best for: Payment teams that need archive and trace on a shared endpoint, with SOC 2 documentation, and predictable per-request billing.

5. dRPC

dRPC routes Plasma traffic across independent node operators behind a load balancer, and labels Plasma "Archive" and "Premium". It is on Plasma's official providers page.

Key features:

• Archive on Plasma
• WebSocket
• Debug and trace on Growth (general, not Plasma-specific), at the same flat 20 CU
• 130+ chains, crypto payment
• Front-running protection and the open-source NodeCore gateway

Pricing:

dRPC's flat rate of $6.00 per 1M calls covers every method, archive and trace included. It is the only partner-listed provider with no free access to Plasma: the 210M CU free tier does not include the chain. Growth advertises 99.99% uptime, with the formal SLA on Enterprise. Because traffic is routed across third-party operators, the node operator serving your Plasma requests is not disclosed.

Best for: Pay-as-you-go teams that want archive and trace at one flat per-request rate with no monthly commitment.

6. Tenderly

Tenderly runs Plasma node RPC at plasma.gateway.tenderly.co (HTTPS and WSS), with Virtual Environments and simulation supported on the chain. Plasma's docs highlight it alongside QuickNode as a hosted RPC option.

Key features:

• Public Plasma gateway (tested working)
• WebSocket on Plasma
• tenderly_simulateTransaction and trace for pre-flight checks
• Virtual Environments (forks) on Plasma
• Debugger, monitoring and alerts

Pricing: Sales-led. Tenderly Free has "No API access", and Tenderly Console is quoted per contract. No public dollar prices were published on 2026-09-28.

Effective $/1M calls: not published, so Tenderly cannot be placed on the same scale as the other providers.

The payments mix averages about 3.15 TU per request. Archive on Plasma is not stated, and debug and trace are not confirmed for the chain. The value is simulation: forking Plasma and replaying a 4337 bundle before you ship.

Best for: Teams that need simulation and forked environments for Plasma payment logic and can negotiate pricing with sales.

7. thirdweb

thirdweb serves Plasma through RPC Edge at https://9745.rpc.thirdweb.com, which works keyless. RPC is bundled into a platform that also covers wallets, account abstraction, gas sponsorship and the Insight indexing API.

Key features:

• Keyless public endpoint
• RPC included in a wider app platform
• Infrastructure SLA 99% on Scale, 99.9% on Pro
• 2,500+ chains

With only 1M calls included, the overage sets the real rate: 100M calls on Growth cost $8.91 per 1M. Overage drops to $6.80 per 1M between 250M and 1B, and $5.40 above 1B. At $8 per 1M plus a $99 platform fee, thirdweb has the highest rate among the providers here that bill one unit per request. Archive, WebSocket and debug support on Plasma are not stated; the public endpoint rejected debug_* and returned "Log response size exceeded" on a 5,000-block USDT0 query.

Best for: Teams already using thirdweb wallets and gas sponsorship who want RPC on the same bill.

8. OnFinality

OnFinality runs a public Plasma endpoint at 5 requests per second and lists "Archive & Trace Access" for the chain.

Key features:

• Public HTTPS and WSS endpoints (5 req/s)
• Archive on Plasma; trace on Growth and above
• Debug transactions on Plasma
• Pay-as-you-go dedicated nodes

OnFinality is cheaper than Dwellir at 100M on list price ($249 vs $299), with archive included. The caveats: Plasma carries higher response-unit weights for intensive methods that are not published, the "+ usage costs" wording is unclear, and Growth's 200 RU/s cap leaves little burst headroom if those weights are high. No formal SLA was found, and crypto payment is Ultimate only.

Best for: Prototyping on the public endpoint and cost-sensitive production that can accept unpublished method weights.

9. Tatum

Tatum added Plasma mainnet and testnet RPC in November 2025 and runs a keyless gateway at plasma-mainnet.gateway.tatum.io.

Key features:

• debug_traceBlock, debug_traceTransaction and trace_block in the Plasma reference
• Dedicated API keys with flat RPS and no credit consumption
• Data API, notifications and MPC wallets
• 100+ blockchain nodes; no crypto payment yet

Credit plans are sold in steps inside those bands, for example 40M credits at $349, 200M at $1,549 and 350M at $2,499. At the 350M step an eth_call (5 credits) costs $35.70 per 1M and a debug or trace call (50 credits) $357 per 1M, so credit plans reach $2,499 for 100M Plasma requests; the dedicated key costs $509 for 50 RPS unmetered. Archive on Plasma is not stated, and the public gateway rejected a historical eth_call at block 1,000,000.

Best for: Teams that want a flat-RPS dedicated key bundled with Tatum's data APIs.

Also Serving Plasma

NOWNodes lists Plasma with archive mode and debug, priced in euros at one request per unit. Its Enterprise plan is EUR 500 for 100M requests, and it is unclear whether the free Start plan includes Plasma.

BlockPI sells prepaid RU packages that expire after 60 or 90 days, and works out to about $217 a month at 100M requests on the payments mix. eth_sendRawTransaction costs 200 RU, which makes write-heavy flows expensive.

Envio HyperRPC is a read-only RPC for historical log and block scans on Plasma. It cannot send transactions, and its pricing was not readable on 2026-09-28.

Provider Comparison

For reference, the official public endpoint rpc.plasma.to measured 25 requests per second and capped eth_getLogs at 10,000 blocks and 20,000 results. Plasma describes it as rate-limited and not for production.

Effective Price per Million Calls

One denominator for every provider: dollars per 1M Plasma calls under the payment mix, at the entry paid plan and at the best self-serve rate, with overage alongside.

What That Costs at Volume

Plans step at fixed tiers, so monthly bills differ from the per-million rates. Same payments mix, list prices.

At 10M requests, Dwellir's $49 ties Chainstack, OnFinality and BlockPI. At 50M and 100M, OnFinality and BlockPI are lower on list price, with the caveats above: OnFinality's Plasma method weights are unpublished, and BlockPI's packages expire and weight sends at 200 RU. Among the other providers, Dwellir is the lowest at both volumes. The Unlimited Node at 50M is $119 if average traffic stays under 25 RPS and $479 for the 100 RPS tier.

Choosing the Right Provider

High-volume payment backend on recent state: Dwellir. Balance reads, receipts, sends and small-window eth_getLogs cost one credit each, Scale runs 5,000 RPS sustained, and the Unlimited Node starts at $119 for 25 RPS. The caveat is archive: the shared endpoint runs full nodes. If you need archive and SOC 2 on the same shared endpoint, Chainstack is the alternative, at $449 for 100M recent-block requests.

Reconciliation and compliance with archive and trace: Chainstack or QuickNode. Chainstack documents both archive and trace_* on Plasma with SOC 2 Type II and ISO 27001. QuickNode has archive, debug and trace plus SOC 1/2 Type 2 and ISO 27001, at 20 credits per call (40 for trace).

Prototyping and testing: OnFinality or Dwellir free tiers, plus Tenderly. OnFinality's public endpoint allows 5 requests per second, and Dwellir's free tier gives 100,000 responses a day at 20 RPS. Tenderly adds simulation and forked Plasma environments if you can get pricing from sales. dRPC has no free Plasma access.

Watching USDT0 Payments With viem

This example subscribes to USDT0 transfers over Dwellir's Plasma WebSocket and checks each one against the finalized block.

import { createPublicClient, defineChain, formatUnits, parseAbiItem, webSocket } from 'viem'

const WSS = 'wss://api-plasma-mainnet.n.dwellir.com/<your-api-key>'

const plasma = defineChain({
  id: 9745,
  name: 'Plasma',
  nativeCurrency: { name: 'XPL', symbol: 'XPL', decimals: 18 },
  rpcUrls: { default: { http: ['https://api-plasma-mainnet.n.dwellir.com/<your-api-key>'], webSocket: [WSS] } },
})

const client = createPublicClient({ chain: plasma, transport: webSocket(WSS) })

const USDT0 = '0xB8CE59FC3717ada4C02eaDF9682A9e934F625ebb'

// Blocks arrive every second, so logs stream in over one subscription
client.watchContractEvent({
  address: USDT0,
  abi: [parseAbiItem('event Transfer(address indexed from, address indexed to, uint256 value)')],
  eventName: 'Transfer',
  onLogs: async (logs) => {
    // `finalized` trails `latest` by about 2 blocks on Plasma
    const finalized = await client.getBlock({ blockTag: 'finalized' })
    for (const log of logs) {
      const settled = log.blockNumber !== null && log.blockNumber <= finalized.number
      // USDT0 has 6 decimals
      console.log(log.args.to, formatUnits(log.args.value ?? 0n, 6), settled ? 'finalized' : 'pending')
    }
  },
})

This example labels each transfer once, on arrival. In production, keep the pending ones in a set and re-check them against finalized on each new block before you settle them.

Getting Started

Before you commit, run your real workload against 2-3 providers. Two Plasma-specific checks matter most: the largest eth_getLogs window each provider returns for USDT0 without hitting a range or result cap, and how many blocks its finalized tag trails latest. Both take minutes to measure.

Free tiers exist at Dwellir, Alchemy, QuickNode (one-month trial), Chainstack, OnFinality and thirdweb (keyless endpoint). dRPC has none for Plasma. For the wider picture, see the Arc and Tempo comparisons, the Stable chain comparison, RPS limits compared, enterprise stablecoin infrastructure and self-hosted vs managed nodes.

Building payments on Plasma? Create a Dwellir account to test the free tier, compare plans, or contact the Dwellir team about a dedicated archive cluster.