Skip to main content

Best account abstraction infra in 2026

Account Abstraction Infra evaluated across published price per user operation, markup on sponsored gas, exit path without the vendor, and verified mainnet chain coverage.

Candide
#1 of 8 · published ranking
Candide
81ChainChoice Score
Cost
Not priced · No comparable price is published
Joint #1 · ordered by name, A to Z
8 compared Ranking-blind · 392 modules checked 2026-09-18Evidence read 2026-08-07Scored under methodology v2026.09.15 (2026-09-16)32 receipts quoted
8account abstraction infra · sorted by chainchoice score
ranked before any payout data is seen
#1 overallcomputed before any payout data is seenOverall
Candide
Candide
Safe-aligned bundler and paymaster with a per-plan ladder, running the LGPL-3.0 Voltaire bundler.
Wins the weighted total without leading any single criterion · level with Pimlico
81ChainChoice Score · first of 8
Catalog strengths
Markup on sponsored gasExit path without the vendorPublished price per user operation
Why it leads
  • Tied with Pimlico on the published total; the tie-break orders them
  • Provider states broad availability
  • Exit without vendor 9/10 (weight 30%)
Evidence
4/4
criteria scored · 4 receipts quoted
Margin
level
over Pimlico, ranked #02
Rank stability
Level
Level with Pimlico on the weighted total from different criterion scores; the criteria alone average 7.9 for both
Tradeoff
−2
Published price per user operation — behind this pool's best
Jurisdiction
Global
no restricted market on record
Score breakdowntick = pool best
Published unit price8/10
Sponsored gas markup9/10
Exit without vendor9/10
Verified chain coverage5/10
Ranking-blind · a guided run tailors this to your size, custody & jurisdiction
#ProviderScoreEvidenceKey strengths
Joint #1 · ordered by name, A to Z
2=
Pimlico
Published price per user operationExit path without the vendor
3−1
Etherspot (Skandha Bundler and Arka Paymaster)
Etherspot (Skandha Bundler and Arka Paymaster)
Markup on sponsored gasPublished price per user operation
4−4
Alchemy Smart Wallets (Gas Manager and Bundler APIs)
Alchemy Smart Wallets (Gas Manager and Bundler APIs)
Published price per user operationExit path without the vendor
5−7
ZeroDev
ZeroDev
Published price per user operationMarkup on sponsored gas
6−1
thirdweb Account Abstraction
thirdweb Account Abstraction
Markup on sponsored gasPublished price per user operation
7−16
Coinbase Developer Platform Paymaster
Markup on sponsored gasPublished price per user operation
8−8
Biconomy
Exit path without the vendorVerified mainnet chain coverage
Ranking-blind — order computed before any payout data is joined
Below the table

How this ranking works

Everything the table draws on continues here: how firm the #1 is, the per-criterion arithmetic behind each score, who pays ChainChoice, and the full guide to choosing.

Direct answer

What is the best account abstraction infra in 2026?

Candide ranks #1 overall for account abstraction infra on ChainChoice. Safe-aligned bundler and paymaster with a per-plan ladder, running the LGPL-3.0 Voltaire bundler. It holds that rank under an affiliate-blind methodology scored across 4 published, weighted criteria — the code that ranks providers physically cannot read affiliate payouts (CI-enforced), so a payout can't move a rank. The verdict re-computes on every fee change, incident, or regulatory action; full reasoning and the audit receipt are below.

Best picks

Best account abstraction infra in 2026

Weighted on the published unit price (30), the markup on sponsored gas (20), whether you can leave without the vendor (30), and verified chain coverage (20). Exit carries as much weight as price because of what the numbers show: summing BundleBear's own bundler table gives 1,230,964,806 user operations across sixteen labelled operators, of which Alchemy alone is 791,397,966 — 64.29% — and Alchemy, Pimlico and Coinbase together are 83.59%. The entire competitive tail here shares 5.40% between them. The infrastructure layer under "decentralised" accounts is more concentrated than most centralised exchanges.
Best overall
Candide
Candide
Safe-aligned bundler and paymaster with a per-plan ladder, running the LGPL-3.0 Voltaire bundler.
Data checked Sep 2026
Safe-aligned bundler and paymaster with a per-plan ladder, running the LGPL-3.0 Voltaire bundler. Strongest on markup on sponsored gas (9/10): Candide states that gas is always passed through at cost with no markup, and publishes the sponsorship service fee as a number: 5,000 sponsored operations a month included on Launch, then $0.01 per operation ($0.02 on Starter, $0.005 on Grow). Weakest on verified mainnet chain coverage (5/10): Candide's supported-networks table lists 14 EVM mainnets with chain IDs; 24 more are an add-on by request, and the $399 Launch plan reaches 7 mainnet chains. Its keyless public endpoint answers on 11 of the 14. Published price: "Launch $399/mo Going to Production ✓ 100,000 UserOps/mo ✓ 7 mainnet chains", "Grow $899/mo Scaling ✓ 500,000 UserOps/mo ✓ All mainnet chains" and "Enterprise $3,000+/mo", with sponsorship priced separately: "Gas is always passed through at cost — we never…
Best for: Markup on sponsored gas — 9/10
Why this score4 published criteria · leads 1 of 4
Published criterionWtScore, and the best hereGap/10Pts
Published price per user operation7.2−287.7
Markup on sponsored gas4.8−195.8
Exit path without the vendor7.2·98.6
Verified mainnet chain coverage4.8−253.2
Σ methodology points25.3/32

Each bar is the score on that criterion’s own 0–10 scale, never rescaled to the pool. The dark line is the best any product here reached on that axis. Wt is the most the criterion can add to the 86-point weighted total. Pts is weight × score × 32; the sum is the methodology score, and each weighted point behind the leader costs 2.6 on the displayed score. how these are weighted

Why it ranks first
Why Candide leads this category right now
Safe-aligned bundler and paymaster with a per-plan ladder, running the LGPL-3.0 Voltaire bundler. Strongest on markup on sponsored gas (9/10): Candide states that gas is always passed through at cost with no markup, and publishes the sponsorship service fee as a number: 5,000 sponsored operations a month included on Launch, then $0.01 per operation ($0.02 on Starter, $0.005 on Grow). Weakest on verified mainnet chain coverage (5/10): Candide's supported-networks table lists 14 EVM mainnets with chain IDs; 24 more are an add-on by request, and the $399 Launch plan reaches 7 mainnet chains. Its keyless public endpoint answers on 11 of the 14. Published price: "Launch $399/mo Going to Production ✓ 100,000 UserOps/mo ✓ 7 mainnet chains", "Grow $899/mo Scaling ✓ 500,000 UserOps/mo ✓ All mainnet chains" and "Enterprise $3,000+/mo", with sponsorship priced separately: "Gas is always passed through at cost — we never…
Best for
Markup on sponsored gas — 9/10
Main tradeoff
Candide is the smallest independently-measured operator in this pool by a wide margin: BundleBear attributes 390,834 UserOps to candide against 122,078,505 for pimlico and 791,397,966 for alchemy (bundlebear.com/erc4337-bundlers/all), roughly three ten-thousandths of the leader. The engineering discipline is real and unusually well documented — "Every minute, on every chain we support, we execute a real signed transaction end to end." — but the SLA is the thing to read twice. Corrected during verification: the SLA quotes belong to docs.candide.dev/wallet/api/reliability.md, not the pricing page, and the two pages describe it differently. The reliability page's table reads "Starter | No SLA", "Launch Prod | No SLA", "Grow | 99% uptime" and "Enterprise | 99.9% uptime, or as specified in contract"; the pricing page's comparison row instead reads "Uptime SLA Best Effort Best Effort 99.0% 99.9% (w/ penalties)". Either way the $399/mo production plan carries no uptime commitment at all.
Verify before signup
"Launch $399/mo Going to Production ✓ 100,000 UserOps/mo ✓ 7 mainnet chains", "Grow $899/mo Scaling ✓ 500,000 UserOps/mo ✓ All mainnet chains" and "Enterprise $3,000+/mo", with sponsorship priced separately: "Gas is always passed through at cost — we never mark up gas prices. The included ops per month are free. After the included amount, a small service fee applies." against "Included Ops 1,000 Ops 5,000 Ops/mo 25,000 Ops/mo Unlimited" and "Fee After Included $0.02/op $0.01/op $0.005/op Zero" (docs.candide.dev/wallet/pricing/, fetched 2026-09-27, HTTP 200, page stamped "Last updated on Jul 24, 2026"; every figure unchanged from 2026-08-07). The free Starter tier is a trial: "2,500 mainnet UserOps — 90 day trial". WORKED EXAMPLE (ChainChoice arithmetic, 2026-09-27, Base): 50,000 unsponsored user operations sit inside the Launch plan's 100,000 UserOps/mo, so the fee is 399.00 US dollars (0.00798 each); 50,000 sponsored operations exceed the 5,000 included sponsorship ops, so the fee is 399.00 + 45,000 x 0.01 = 849.00 US dollars, or 0.01698 each, the most expensive sponsored service fee in the pool.
Recommendation summary
What should decide this category
What does one user operation cost, and what does sponsoring the gas add?
If this vendor stops, can your users still get their transactions included?
Is the chain list you were sold the chain list the bundler actually answers on?
Quick picks
Strong options in this category
Start with the lead choice first, then use the shortlist only if you still need a challenger or stronger fit for a specific setup.
Best overall
Candide
Candide
Safe-aligned bundler and paymaster with a per-plan ladder, running the LGPL-3.0 Voltaire bundler.
Safe-aligned bundler and paymaster with a per-plan ladder, running the LGPL-3.0 Voltaire bundler. Strongest on markup on sponsored gas (9/10): Candide states that gas is always passed through at cost with no markup, and publishes the sponsorship service fee as a number: 5,000 sponsored operations a month included on Launch, then $0.01 per operation ($0.02 on Starter, $0.005 on Grow). Weakest on verified mainnet chain coverage (5/10): Candide's supported-networks table lists 14 EVM mainnets with chain IDs; 24 more are an add-on by request, and the $399 Launch plan reaches 7 mainnet chains. Its keyless public endpoint answers on 11 of the 14. Published price: "Launch $399/mo Going to Production ✓ 100,000 UserOps/mo ✓ 7 mainnet chains", "Grow $899/mo Scaling ✓ 500,000 UserOps/mo ✓ All mainnet chains" and "Enterprise $3,000+/mo", with sponsorship priced separately: "Gas is always passed through at cost — we never…
Best for: Markup on sponsored gas — 9/10
Published price per user operation · 30%
8/10
Markup on sponsored gas · 20%
9/10
Exit path without the vendor · 30%
9/10
Verified mainnet chain coverage · 20%
5/10
Joint first
Level with Candide on the weighted total from different criterion scores; the criteria alone average 7.9 for both
Joint #1 · ordered by name, A to Z
Pimlico
ERC-4337 bundler and verifying paymaster sold on metered API credits, with the Alto bundler released under GPL-3.0.
ERC-4337 bundler and verifying paymaster sold on metered API credits, with the Alto bundler released under GPL-3.0. Strongest on published price per user operation (10/10): Pimlico's docs pricing page puts a per-method credit table, the price of $1 per 100,000 credits and its simulation assumption (750 credits per unsponsored operation, 1,050 sponsored) on one page, reproducing its headline of about $0.0075 per operation ($0.0105 sponsored). Weakest on markup on sponsored gas (4/10): Pimlico bills sponsored mainnet gas at actual cost plus a 10% surcharge, with no surcharge on testnets; its docs show the formula as actual gas cost times 1.1. Published price: Pay-as-you-go on pimlico.io/pricing (fetched 2026-09-27, HTTP 200): "$0 per month, card required", "10,000,000 credits / month included", "Then $1 per additional 100,000 credits", "~$0.0075 per user operation (~$0.0105 sponsored)" and "10% surcharge on…
Best for: Published price per user operation — 10/10
Published price per user operation · 30%
10/10
Markup on sponsored gas · 20%
4/10
Exit path without the vendor · 30%
9/10
Verified mainnet chain coverage · 20%
7/10
Quick pick
Etherspot (Skandha Bundler and Arka Paymaster)
Etherspot (Skandha Bundler and Arka Paymaster)
Prepaid-credit bundler and paymaster with both components released under MIT and gas billed at network cost.
Prepaid-credit bundler and paymaster with both components released under MIT and gas billed at network cost. Strongest on markup on sponsored gas (10/10): Etherspot states that gas fees are charged separately at network cost with no markup; its only sponsorship charge is the published credit cost of pm_sponsorUserOperation, 700 credits at $1 per 250,000 credits. Weakest on verified mainnet chain coverage (5/10): Etherspot's chains page lists 17 mainnets, each with a chain ID and bundler URL, plus 13 testnets, and claims no larger headline count. Published price: "Pay-As-You-Go $1 per 250,000 credits Pay only for what your app uses. No subscriptions required", with "Gas fees charged separately at network cost (no markup)" and "Minimum initial top-up: $10 in prepaid credits.
Best for: Markup on sponsored gas — 10/10
Published price per user operation · 30%
8/10
Markup on sponsored gas · 20%
10/10
Exit path without the vendor · 30%
8/10
Verified mainnet chain coverage · 20%
5/10
Quick pick
Alchemy Smart Wallets (Gas Manager and Bundler APIs)
Alchemy Smart Wallets (Gas Manager and Bundler APIs)
Bundler and gas manager sold as compute units inside the general Alchemy node platform, running the Rust Rundler bundler.
Bundler and gas manager sold as compute units inside the general Alchemy node platform, running the Rust Rundler bundler. Strongest on published price per user operation (8/10): Alchemy meters bundler and paymaster calls in compute units: its compute-unit table prices each method (eth_sendUserOperation 1,000 CU, 3,000 with a sponsorship policy) and its pricing page sets $0.525 per million CU, so the cost per operation closes across two pages. Weakest on markup on sponsored gas (6/10): Alchemy charges an 8% admin fee on sponsored gas under Pay as You Go, stated on its pricing page and in the Gas Manager FAQ, and requires the gas to be pre-loaded. Published price: Pay as You Go on alchemy.com/pricing (fetched 2026-09-27, HTTP 200) is metered in compute units, not operations: "$0.525/1M CU", with the gas sponsorship rows reading "Admin fee — 8% Custom" and "Limit — Pre-load funds Custom" (Free, Pay as You Go…
Best for: Published price per user operation — 8/10
Published price per user operation · 30%
8/10
Markup on sponsored gas · 20%
6/10
Exit path without the vendor · 30%
7/10
Verified mainnet chain coverage · 20%
7/10
Frequently asked
Questions people ask before choosing account abstraction infra
How concentrated is this layer really?
More than the word "decentralised" in front of it suggests. Re-derived from the raw page data at bundlebear.com/erc4337-bundlers/all on 2026-08-07: 1,230,964,806 user operations across sixteen labelled operators, with Alchemy at 791,397,966 (64.29%) and Alchemy, Pimlico and Coinbase together at 1,028,954,948 (83.59%). A further 130,710,666 (10.62%) are attributed to "Unknown". Everything else in this comparison — thirdweb, Biconomy, Particle, Gelato, Stackup, Candide, Etherspot — shares 66,432,774 between them, or 5.40%. That is why the ability to leave carries 30% of the weight here rather than being a footnote.
Is the advertised chain count the real one?
Not always, and one of them is testable without an API key. thirdweb's pricing page advertises "Chains Supported 145+"; its bundler is addressed per chain and distinguishes a missing key from an unsupported chain, so coverage can be probed keylessly. Of 24 chain IDs probed on 2026-08-07, nine answered "Invalid chain" — Zora, Blast, Mode, Mantle, Unichain, Sonic, HyperEVM, Ink and Soneium — and every one of those nine carries live, byte-identical EntryPoint v0.6 and v0.7 contracts read in the same session. Pimlico publishes three incompatible counts: "100+ chains" on the pricing page, "all 25+ mainnet chains" in the docs pricing table, and an enumerated reference that counts 43 mainnets and 42 testnets, or 85 named networks.
What is a paymaster stake and why should I check it?
Under ERC-4337 reputation rules an unstaked paymaster faces tighter throttling in the canonical mempool — exactly the condition in which you would need a fallback to work. It is checkable directly: calling EntryPoint.getDepositInfo on 2026-08-07 shows Pimlico's ERC-20 paymaster staked on Base for v0.6 and v0.7 (1.9708 and 1.4701 ETH, 1,209,600-second unstake delay) but staked=false with stake=0 for v0.8; on Ethereum mainnet all three are unstaked despite holding deposits of 1.0663, 1.2600 and 1.3893 ETH; on Arbitrum v0.7 is staked and v0.8 is not. None of that appears on any Pimlico page.
Free advisorNo signup needed

Still unsure? Get your best account abstraction infra pick

Answer a few quick questions and get one clear recommendation based on how you actually plan to use crypto — then review the evidence before deciding.

No signupFree first pass · PrivateNo spam · No account

Not financial advice · Independent · Always do your own research

Browse this network
How this ranking is built
Reviewed on published price per user operation, markup on sponsored gas, exit path without the vendor, and verified mainnet chain coverage.
Data checked Sep 2026 · Independent rankings · We show our work
Not financial advice · For informational purposes only · Always do your own research
ChainChoice
Compare onchain. A brighter future.
© 2026 ChainChoice. All rights reserved.
System statusFeeds last refreshed 9 days ago
Built from commit ba0993b · rankings.json sha256 e1eb24980abb
ChainChoice · The decision layer for crypto · Not financial adviceEducational analysis, not investment advice. Affiliate links may contribute to operations but never alter rankings.

ChainChoice provides informational content only. Nothing on this site constitutes financial, investment, legal, or tax advice. Always do your own research and consult a qualified professional before making financial decisions.