Embedded Wallet Providers Compared: Privy, Web3Auth, Turnkey, Para

Embedded Wallet Providers Compared: Privy, Web3Auth, Turnkey, Para

Reviewed by Jordan Paul · November 18, 2026 · 16 min read

Quick Answer

No embedded wallet provider wins every case. Rated on documented features and published pricing checked on 18 November 2026, Para ranks first at 7.9 out of 10, ahead of Openfort, thirdweb and Turnkey. At 10,000 monthly active wallets, published plans cost from $50 to $2,000 a month, before gas and SMS fees. Turnkey prices per signature.

LiveBTC—·ETH—·Gas—via CoinGecko

The best embedded wallet provider depends on whether you want the lowest bill, the most clearly documented custody model or the simplest login. Rated on the criteria below, Para scores highest among the embedded wallet providers at 7.9 out of 10, followed by Openfort, thirdweb, Turnkey, Crossmint, Web3Auth, Privy and Magic. This guide compares eight embedded wallet providers on documented features and published pricing, checked on 18 November 2026. It is an editorial synthesis, not a hands-on benchmark.

An embedded wallet is a wallet created inside an app and tied to a login instead of a seed phrase. Providers differ in how they protect the key and in what they count when they bill you. This guide explains both, then ranks the providers. If wallets are new to you, start with what a crypto wallet is.

How We Ranked These Embedded Wallet Providers

We rated each provider on six criteria, with the same weights for every entry. Scores run from 0 to 10 per criterion and are combined into a weighted total out of 10. Rank is by merit only. Nobody paid for a position, and there are no affiliate links in this guide.

  • Published pricing and cost (30%): how clear the pricing page is, and the monthly cost at 1,000, 10,000 and 100,000 active wallets.
  • Key custody and export (20%): how the vendor's own docs describe key storage, who can sign, and whether users can leave.
  • Auth and feature coverage (15%): login methods and wallet features listed on the pricing page and docs.
  • Developer experience (15%): SDK and docs coverage, and how visible and frequent the dated release history is.
  • Policy and compliance controls (10%): transaction policies, audits and certifications the vendor itself states.
  • Ownership and lock-in (10%): documented exit routes, plus the vendor's own statements about who owns it.

For every embedded wallet provider here, evidence is limited to vendor pricing pages, documentation, changelogs and company announcements, read on 18 November 2026. Where a vendor's page did not answer a question, the entry says so and gives no score credit. Acquisition claims appear only when a company's own announcement confirms them.

RankProviderPricing (30%)Custody (20%)Coverage (15%)Dev exp. (15%)Policy (10%)Lock-in (10%)Score
---------------------------
1Para7979887.9
2Openfort8888687.8
3thirdweb8788687.6
4Turnkey6878977.3
5Crossmint7876877.2
6Web3Auth7777556.6
7Privy4779856.3
8Magic6766566.1

Scores are rounded to one decimal, so close totals are near ties. Dynamic is not in the table. See "Also considered, not scored" below.

1. Para — Best for documented MPC custody

Best for: teams that want an embedded wallet with a clearly specified 2-of-2 multi-party computation (MPC) design and a steady release cadence.

Para's docs describe a 2-of-2 MPC architecture in which the private key is split between the user's device and Para's cloud hardware security modules. Para's share cannot produce a valid signature alone, and the docs state the full key is never assembled during generation, signing or recovery.

  • Newest version: v3.28.0 on 13 November 2026, per Para's dated changelog. The two releases before it landed on 9 and 5 November.
  • Pricing: Free up to 1,200 MAU, Starter $200 a month (2,500 MAU), Growth $500 (10,000 MAU) and Scale $1,000 (25,000 MAU), as of 18 November 2026. Overage is $0.06, $0.05 and $0.04 per extra MAU on the three paid plans.
  • Custody: the docs call the model self-custodial and non-custodial, state SOC 2 Type II compliance, and say users can export their Para share and sign without Para's servers.
  • Features: passkeys, phone login, pregenerated wallets, wallet recovery and multichain support from Starter.

Limitations: the Free plan lists no features, REST API limits apply (30 requests a minute on Free), and Enterprise pricing is unpublished. Growth beyond 25,000 MAU is not priced beyond the $0.04 overage figure on the page.

2. Openfort — Best for the lowest cost at scale

Best for: apps with many active wallets that want published operation pricing and an exit route.

Openfort bills its embedded wallet product by operations rather than users. Creating a wallet, broadcasting a transaction, signing a payload, sending a webhook, evaluating a policy and exporting a key each count as one operation. Its own key-management article says embedded wallets use Shamir secret sharing, with shares held in its Shield service and in the user's account.

  • Newest version: @openfort/openfort-js 2.6.4, released 23 October on GitHub (the page shows no year).
  • Pricing: Free $0 (2,000 operations), Growth $99 (25,000), Pro $249 (100,000) and Scale $599 (500,000), as of 18 November 2026. Overage runs from $0.01 down to $0.004 per operation.
  • Custody: backend wallets run in a trusted execution environment (TEE), and the signer is open source and self-hostable. Openfort lists audits by Cure53 (September 2024, Shamir) and Quantstamp (October 2025, key management).
  • Features: social, one-time-passcode, anonymous and external sign-in, account abstraction, gas sponsorship, private key export and transaction policies on every plan.

Limitations: gas sponsorship carries a 10% paymaster surcharge on Free and Growth and 5% on Pro and Scale. Openfort's own article says it does not hold SOC 2 Type II, and operation counts can climb faster than signatures.

3. thirdweb — Best for one bundle of wallets and infrastructure

Best for: teams that also want server wallets, gas sponsorship, RPC and payments from one vendor.

thirdweb prices its embedded wallets per monthly active wallet on top of a plan, and the same page prices RPC, storage, webhooks and an x402 facilitator. Its FAQ says the system is designed to be non-custodial and that a breach of thirdweb or a developer's database would expose only encrypted keys.

  • Newest version: thirdweb 5.122.1, released 30 October on GitHub (no year shown). The previous release was 12 October.
  • Pricing: Growth $99 a month, Scale $499 and Pro $1,499 or more. Wallets are free for the first 1,000, then $0.015 each up to 100,000 and $0.01 up to 1,000,000, as of 18 November 2026.
  • Export: users can export a key from the Connect modal, but programmatic export is not possible, and developers can hide the option.
  • Continuity: thirdweb states it would run its user service for at least 12 months after announcing any shutdown.

Limitations: the FAQ says an enclave protects assets but does not explain it or mention key shares. Fees add up: 2.5% on mainnet gas sponsorship, 0.3% on bridge swaps and x402, and SMS delivery is listed without a price.

4. Turnkey — Best for enclave-based policy control

Best for: products that need policy-gated signing and want to bill per signature.

Turnkey, which builds embedded wallet infrastructure, runs key generation, signing and its policy engine inside secure enclaves on AWS, with no persistent storage, no interactive access and no external networking. Its docs say only encrypted key material is stored and that it is decrypted inside an enclave when the owner authenticates, so raw private keys are never exposed to Turnkey.

  • Newest version: JS SDK v2026.10.4, published 24 October on GitHub. The page shows no year, though the tag implies 2026.
  • Pricing: Pay as You Go has 25 free signatures, then $0.10 each. Pro is $99 a month at $0.05 per signature, as of 18 November 2026. Broadcasting costs $0.20 (Pay as You Go) or $0.10 (Pro) per transaction.
  • Limits: wallets are capped at 1,000 on Pay as You Go and 2,000 on Pro. Policies are capped at 5 per organization, or 250 on Enterprise.
  • Gas: sponsorship up to $10 a month on Pay as You Go and $25 on Pro.

Limitations: the wallet caps push any larger app to a custom quote, and the pricing page does not say how the Pro minimum interacts with per-signature fees. The pages we read did not address key export.

5. Crossmint — Best for choosing a custody model per wallet

Best for: teams that need non-custodial, custodial or hybrid wallets, including agent wallets.

Crossmint's embedded wallet custody docs describe three models decided by who holds the signers: non-custodial, custodial and hybrid. For non-custodial setups the docs say Crossmint has no ability to authorize transactions, and they warn that a user-controlled signer with an organization-controlled recovery method is hybrid, not non-custodial.

  • Newest version: no GitHub releases are published. Its docs changelog lists 3 November 2026 as the newest entry: Crossmint Auth is restricted to staging, so production calls return an error by default. The newest wallet feature entry is 11 February 2026 (transfer webhooks).
  • Pricing: the Developer plan is free for the first 5,000 monthly active wallets, then $0.01 each, as of 18 November 2026. Gas sponsorship is capped at $50.
  • Fees: the page says gas and OTP costs for SMS, WhatsApp and email are billed separately and publishes no rates.
  • Custom plan: adds a policy engine, multi-approval and AML screening, at unpublished prices.

Limitations: no fixed plans between free and Custom, and the production-auth change means you must bring another login provider.

6. Web3Auth — Best for social login with share-based keys

Best for: apps that want social and email login with a tiered, low-cost plan ladder.

Web3Auth's embedded wallet docs now sit under MetaMask. They say the wallet is split into multiple key shares using Shamir's Secret Sharing, and that no single party, including MetaMask or the login provider, can control the wallet from one share. Consensys announced the acquisition on its own site on 2 June 2025.

  • Newest version: v11.4.3 of web3auth-web, listed 12 October on GitHub (no year shown). Docs show no release notes.
  • Pricing: Base is free for 1,000 monthly active wallets (MAW). Growth is $69 for 3,000, Scale is $399 for 10,000, and Enterprise is custom, as of 18 November 2026. Overage is $0.05, $0.045 and $0.04 per extra MAW.
  • Plan gating: passkeys and native account abstraction start at Growth, pre-generated wallets at Scale.
  • MPC: the threshold-signature Core Kit, which avoids key reconstruction, is listed on Enterprise only.

Limitations: the pages we read did not cover key export. The pricing page stays on web3auth.io while docs moved to a MetaMask domain, so confirm which terms apply.

7. Privy — Best for documented TEE architecture

Best for: teams that value a detailed public architecture description and weekly product releases.

Privy's embedded wallet architecture page says a trusted execution environment generates the wallet entropy, which is split by Shamir's secret sharing into two shares. The enclave holds one, encrypted. Privy stores the other, the auth share, encrypted and released only on valid credentials. The page never uses the word non-custodial and does not describe key export.

  • Newest version: the product-updates changelog's newest entry on or before 18 November is 16 October 2026 (Earn deposit automations).
  • Pricing: Developer is free for 0 to 499 MAU, Core $299 for 500 to 2,499 and Scale $499 for 2,500 to 9,999, as of 18 November 2026. Above 10,000 MAU or 50,000 signatures, pay-as-you-go applies: $2,000 base, $0.05 per extra MAU and $0.01 per extra signature.
  • Included: 50,000 signatures and $1 million in transaction volume each month.

Limitations: the cost cliff is steep, with a $2,000 base at 10,000 MAU. SMS and gas sponsorship are listed without prices. The pages we read name no parent company, so no ownership claim is made.

8. Magic — Best for a free start with signature-based billing

Best for: small apps that want a free tier with 50,000 signatures included.

Magic's docs describe non-custodial embedded wallets for each user, secured by what it calls a patent-pending TEE Key Management System, and list key export as a feature. The pages we read do not explain how keys are split or stored.

  • Newest version: magic-sdk 33.13.0, listed 1 September on GitHub (no year shown), the oldest release date among the providers that publish GitHub releases.
  • Pricing: Developer is $0 with 1,000 MAW and 50,000 signatures. Startup is $99 with 2,500 MAW and five team seats. Overage is $0.045 or $0.04 per MAW and $0.005 per signature, as of 18 November 2026.
  • Billing: signatures are pooled per team across apps and wallets, and annual billing is 10% off.

Limitations: per-signature fees make heavy signers costly, and the page shows no Enterprise figures. Public detail on custody and policy controls is thinner than for peers.

Which Embedded Wallet Provider Is Cheapest?

At 10,000 monthly active wallets, Crossmint's published plan is the cheapest at about $50 a month, followed by thirdweb ($234) and Openfort ($249). Privy is the most expensive at $2,000. Turnkey cannot be priced at that size from published plans. These totals exclude gas, SMS and passcode fees.

Worked cost comparison

We assume each active wallet makes five signatures a month, uses no gas sponsorship, and sends no SMS. Costs are monthly, from each pricing page on 18 November 2026, choosing the cheapest published plan that fits. Wallet creation and webhooks are ignored, which flatters Openfort.

Provider1,000 wallets10,000 wallets100,000 wallets
------------
Crossmint$0$50$950
thirdweb$99$234$1,584
Openfort$99$249$599
Web3Auth$0$384$3,999
Magic$0$399$6,249
Para$0$500$4,000
Privy$299$2,000$11,000
Turnkey$250 or $349quote neededquote needed

The arithmetic, with signatures at five per wallet (5,000, 50,000 and 500,000 a month):

  • Crossmint: 10,000 minus 5,000 free is 5,000 wallets at $0.01, which is $50. At 100,000 it is 95,000 times $0.01, or $950.
  • thirdweb: $99 plan plus 9,000 paid wallets at $0.015 ($135) is $234. At 100,000 it is $99 plus 99,000 at $0.015 ($1,485), or $1,584. We assume the plan fee and wallet charges add.
  • Openfort: 5,000 operations needs Growth ($99). 50,000 operations on Pro is $249, against $299 on Growth. 500,000 fits Scale's allowance, so $599.
  • Web3Auth: Growth at 10,000 is $69 plus 7,000 at $0.045 ($315), or $384. Scale at 100,000 is $399 plus 90,000 at $0.04 ($3,600), or $3,999.
  • Magic: Startup at 10,000 is $99 plus 7,500 at $0.04 ($300), or $399. At 100,000 it is $99 plus 97,500 at $0.04 ($3,900) plus 450,000 extra signatures at $0.005 ($2,250), or $6,249.
  • Para: Growth covers 10,000 at $500. At 100,000, Scale is $1,000 plus 75,000 at $0.04 ($3,000), or $4,000. We assume that overage rate continues past 25,000 MAU.
  • Privy: $299 at 1,000. At 10,000 we treat the $2,000 base as the start of pay-as-you-go. At 100,000 it is $2,000 plus 90,000 at $0.05 ($4,500) plus 450,000 extra signatures at $0.01 ($4,500), or $11,000.
  • Turnkey: 5,000 signatures on Pro is $250 if the $99 is a minimum, or $349 if additive. Wallet caps of 2,000 rule out larger sizes.

Change the signature assumption and the ranking moves. Signature-billed plans (Turnkey, Magic) get cheaper with fewer signatures and costlier with more, while user-billed plans (Para, Crossmint) do not.

How Do Embedded Wallet Providers Price Their Plans?

Embedded wallet providers bill in four ways: by monthly active users or wallets (Privy, Web3Auth, Para, Crossmint, thirdweb), by signatures (Turnkey, with Magic adding them), by operations (Openfort), and by per-transaction fees. Most embedded wallet vendors add separate charges for gas sponsorship, SMS and one-time passcodes, so a headline price is rarely the total.

Definitions differ between embedded wallet providers. Para counts a user who is authenticated to or created in your app. Openfort counts any wallet creation, signature, webhook or policy check. Because gas and OTP rates are often unpublished, ask for them in writing. For the gas side of wallets, see our guides to ERC-4337 account abstraction and what account abstraction is.

Are Embedded Wallets Really Non-Custodial?

Embedded wallets are non-custodial only if the provider cannot sign or reconstruct the key without the user's credential. Embedded wallet vendors use different mechanisms: enclaves, key shares and MPC. The label is a design claim, not a legal conclusion, and recovery methods can change it.

ProviderMechanism stated in its docsNon-custodial claim in the pages we read
---------
Para2-of-2 MPC, device share plus HSM shareYes, "self-custodial / non-custodial"
OpenfortShamir shares; TEE for backend walletsDocumented non-custodial configuration; backend wallets can be custodial
thirdwebEnclave, details not given"Designed to be non-custodial"
TurnkeySecure enclaves, ciphertext onlyOwner remains owner of keys
CrossmintSigners chosen per walletYes for non-custodial setup; hybrid otherwise
Web3AuthShamir shares across factorsNo single party holds the full key
PrivyTEE plus 2-of-2 Shamir sharesTerm not used
MagicTEE Key Management SystemYes, per its overview page

For comparison, self-custody on dedicated hardware is covered in our hardware wallet guide.

Privy vs Turnkey vs Para: Which Should You Choose?

Among these embedded wallet providers, choose Privy if its detailed public architecture and weekly releases matter more than cost. Choose Turnkey if you need enclave-enforced policies and can live with per-signature billing. Choose Para if you want documented 2-of-2 MPC and steady pricing tiers. At 10,000 wallets, Para costs $500, Privy $2,000 and Turnkey needs a quote.

Web3Auth sits between these embedded wallet providers on price and ranks sixth, mainly because the pages we read left export and compliance questions open. All four are credible, so pilot your real signing volume before deciding.

Also Considered, Not Scored

Dynamic. Its pricing page at dynamic.xyz/pricing returns a 301 redirect to fireblocks.com/dynamic-pricing. That page failed to load twice on 18 November 2026 with a network error, so no plan, price or limit was verified. We make no ownership claim about Dynamic either. Under our rules, an entry whose price cannot be verified is not scored. Check it yourself in a browser.

Which Should You Choose?

Lowest bill at scale: Openfort ($599 at 100,000 wallets in our model) or Crossmint ($950), if you accept that Crossmint publishes no gas or OTP rates and that Crossmint Auth is staging-only.

Most clearly documented MPC custody for an embedded wallet: Para, with a 2-of-2 design, SOC 2 Type II statement and share export.

Policy-gated signing in an embedded wallet: Turnkey, with enclave-enforced policies and per-signature billing, though its self-serve wallet caps are low.

One vendor for wallets plus infrastructure: thirdweb, which prices wallets, RPC, storage and an x402 facilitator on one page. For agent payments, see how x402 lets AI agents pay.

Social login on a low-cost ladder: Web3Auth, starting free at 1,000 wallets.

Weekly release cadence and a public architecture page: Privy, if you can budget for it.

Small app, free embedded wallet start, signatures included: Magic, up to 50,000 signatures.

None of this is investment advice or a recommendation to hold any crypto asset. Before you commit to an embedded wallet provider, request written terms on gas sponsorship, OTP costs, key export and what happens to user keys if the vendor shuts down or changes ownership.

Conclusion

Embedded wallet pricing is not comparable until you fix a usage model, and custody labels are not comparable until you read the mechanism. On 18 November 2026 the published monthly cost at 10,000 wallets varied 40-fold. The cheapest plan is not always the most clearly documented custody model.

This is an editorial synthesis of documentation, public data, and community reports; see our [methodology](/methodology). Verify current details with each project.

Sources

Pricing pages, documentation and announcements read on 18 November 2026:

Key Takeaways

  • On 18 November 2026 the published monthly cost of 10,000 active wallets ranged from $50 (Crossmint) to $2,000 (Privy), a 40-fold spread, before gas, SMS and one-time-passcode fees.
  • Para ranks first at 7.9 out of 10 and Openfort second at 7.8, on a fixed 30/20/15/15/10/10 weighting applied to all eight providers.
  • Providers name different key-protection mechanisms: Privy and Web3Auth split keys into Shamir shares, Turnkey and Magic use enclaves, and Para uses 2-of-2 MPC. Only some pages use the word non-custodial.
  • Providers bill four different units: monthly active users or wallets, signatures, operations and per-transaction fees, so a headline price cannot be compared without a stated usage model.
  • Turnkey's self-serve plans cap wallets at 1,000 (Pay as You Go) and 2,000 (Pro), so a 10,000-wallet app needs a custom Enterprise quote.
  • Dynamic is not scored: its pricing page redirects to a Fireblocks URL that did not load on 18 November 2026, so no price could be verified.
  • Of the ownership claims we checked, only Consensys's purchase of Web3Auth (announced 2 June 2025 on consensys.com) could be confirmed on a company page; the others are left out.

Frequently Asked Questions

What is an embedded wallet?

An embedded wallet is a crypto wallet created inside an app, usually when a user signs in with email, a social account or a passkey. The user never handles a seed phrase. A provider generates and protects the key, using key shares, enclaves or multi-party computation, and the app requests signatures through an SDK.

Which embedded wallet provider is cheapest?

At 10,000 monthly active wallets and five signatures each, Crossmint's published plan costs about $50 a month, thirdweb about $234 and Openfort about $249. That excludes gas, SMS and one-time-passcode fees, which several providers bill separately. Turnkey bills per signature and cannot be priced at that size without a quote.

Are embedded wallets really non-custodial?

It depends on who can reconstruct the key. Turnkey and Para state that the provider cannot sign alone, and Crossmint says its non-custodial configuration leaves signing with the user. Privy's architecture page never uses the word non-custodial. Read each provider's custody model and recovery method before relying on the label.

Can users export their keys from an embedded wallet?

Some can. Openfort lists private key export on every plan, thirdweb lets users export through its Connect modal but not programmatically, and Para says users can export their Para share and sign without its servers. Privy's architecture page and Turnkey's enclave pages do not describe export, so confirm it before you commit.

What does MAU or MAW mean in embedded wallet pricing?

MAU is monthly active users and MAW is monthly active wallets. Both count distinct users or wallets that did something in the month, but each vendor defines activity its own way. Para counts a user authenticated to or created in your app. Turnkey and Openfort skip this unit and bill signatures or operations.

Is the Web3Auth embedded wallet now part of MetaMask?

Consensys announced on 2 June 2025 that it had acquired Web3Auth, and Web3Auth's documentation now lives at docs.metamask.io/embedded-wallets. Its pricing page remains on web3auth.io with plans from free to custom. Check current terms and roadmap with the vendor before building on it.

Reviewed by

Jordan Paul

CTO, Appear

Reviewed for technical accuracy on November 18, 2026

Jordan Paul is a chief technology officer and enterprise solutions architect who has built production blockchain, machine-learning and real-time tracking systems, with clients including Tesla, Medallia and Atlassian. He is CTO at Appear and at TZRO.ai, and previously served as CTO of Xucre Wallet, Smart Campaigns and REC Protocol. His listed top skills are smart contracts, Ethereum and blockchain, and he holds Salesforce architect certifications.