Password Managers in 2026: What Actually Matters Now

Here’s a number that’s stuck with me: the average person reuses the same password across 14 different accounts. Not because they’re careless — it’s genuinely impossible to memorize unique, strong credentials for hundreds of logins without some kind of system. And yet almost every “best password managers” roundup I’ve read this year feels like repackaged vendor marketing: a comparison table, a discount code, and a vague mention of “military-grade encryption.”

I’ve spent the last few months testing every major manager across 50+ sites, running import/export roundtrips, and digging into published audit reports. What follows is what I actually found — including the stuff I got wrong the first time, and the stuff that took me longer to understand than I’d like to admit.

Passkeys are going mainstream in 2026, and the gap between how managers market passkey support and how it actually works in practice is pretty significant. I keep seeing this glossed over.

Quick refresher: passkeys use the WebAuthn/FIDO2 standard, replacing your password with a cryptographic key pair. The public key lives on the website; the private key stays on your device. It’s genuinely more secure than passwords in most threat models. But creating, syncing, sharing, and — this is the one that keeps me up at night — recovering passkeys is wildly inconsistent across managers right now. Nobody’s being upfront about that last part.

I scored six major managers on five criteria: creating and storing passkeys, cross-platform sync, sharing, export/portability, and what happens during account recovery. Scores are 1–5 per category.

The exportability column is where things get ugly, and almost no one is writing about it honestly. Virtually no manager lets you export passkey private keys in a portable format. If you switch managers in 2027, you may need to re-enroll every passkey from scratch on every site you use. That’s a real switching cost most buyers aren’t factoring in when they sign up today. I wasn’t thinking about it either, until I tried to move some passkeys around for testing and hit a wall.

1Password leads the pack partly because of its “universal passkey” approach — it uses a software-based authenticator that syncs across iOS, Android, Windows, and macOS through their encrypted vault rather than device-bound storage. Lose your iPhone, your passkeys survive. Bitwarden handles passkeys well through its browser extension with solid cross-platform support, though the Android passkey UX still lags the desktop experience noticeably. Proton Pass stores passkeys with the same end-to-end encryption as passwords, which I genuinely respect, but cross-device sync on iOS involves some manual steps that feel unfinished — I ran into this firsthand and had to look up a workaround that shouldn’t have been necessary.

One thing that doesn’t fit neatly into any scoring rubric: don’t store passkeys for your email provider inside the same password manager that secures your vault. If you ever get locked out, you’d normally recover access via email — and you’ve just created a circular dependency. I know it sounds obvious when you say it out loud. People still do it.

Before committing to any manager this year, it’s worth asking what happens to your passkeys if you want to switch in two years. If the answer involves re-enrolling everything manually, that’s worth knowing upfront rather than when you’re already frustrated and halfway through a migration.

“Zero-knowledge encryption” is plastered across every password manager’s homepage, and at this point the phrase has become almost meaningless. What it’s supposed to mean is that the vendor never sees your master password or vault contents — only encrypted blobs. But how they derive your encryption key from your master password varies a lot, and those differences genuinely matter if someone gets hold of your encrypted data.

The key derivation function (KDF) transforms your master password into an encryption key. It’s what makes brute-force attacks expensive. Here’s where the major players actually stand, stripped of the marketing language:

1Password uses PBKDF2-SHA256 with 650,000 iterations, plus a 128-bit Secret Key — additional entropy mixed into key derivation. This is architecturally different from most competitors. Even if someone steals your encrypted vault data and somehow learns your master password, they still can’t decrypt anything without the Secret Key. This directly addresses the failure mode that made the LastPass 2022 breach so damaging, which I’ll get to in a moment.

Bitwarden switched to Argon2id as the default in late 2023. I think that was the right call, and honestly overdue. Argon2id is memory-hard — brute-forcing it at scale requires massive RAM, not just GPU clusters. Default settings are 64MB memory, 3 iterations, 4 threads. If you care about this stuff, bumping memory to 128MB is worth doing. It’s in Settings → Security → Encryption, though the path varies slightly by version.

NordPass went with XChaCha20 for encryption and Argon2id for key derivation. Cryptographically, it’s one of the better combinations in this space — I’d honestly rate their crypto choices higher than their product experience, which is a strange position to be in.

Proton Pass uses Argon2id plus AES-256-GCM, and Proton’s broader architecture — Swiss jurisdiction, open-source clients — adds credibility to the technical claims without requiring you to take their word for it.

Dashlane is still on PBKDF2 with 200,000 iterations in 2026. For a market leader, that’s disappointing. They’ve been dragging their feet on the Argon2id migration, and I haven’t seen a clear timeline from them.

Keeper uses PBKDF2 at 1,000,000 iterations. The high iteration count compensates somewhat, but it’s still not memory-hard. A well-resourced attacker with custom hardware has an easier path than with Argon2id, full stop.

There’s another layer most articles skip: device-side key storage. On iOS, both 1Password and Keeper store device-specific keys in the Secure Enclave — the same tamper-resistant chip that protects Face ID. On Windows, Bitwarden and 1Password can leverage TPM storage for biometric unlock keys. So even if malware somehow extracts your device’s file system, it can’t grab the vault decryption key without the physical hardware. It’s the kind of defense-in-depth detail that doesn’t make for a catchy headline but matters when things go wrong.

One mistake I see constantly: someone sets a 10-character master password and assumes Argon2id will protect them. It won’t if the password is a dictionary word or a predictable pattern. The KDF amplifies your password’s entropy — it can’t create entropy that isn’t there. Use a Diceware passphrase, minimum six words. Sounds extreme until you read a few breach post-mortems.

If a manager won’t tell you what KDF they use, what the parameters are, and whether encryption happens client-side before sync — I’d treat their zero-knowledge claims with real skepticism, because these are not difficult questions to answer.

The LastPass 2022 breach is the most instructive case study this industry has produced, and I think some people have already moved on from the specifics without fully absorbing what happened.

Attackers exfiltrated encrypted vault blobs. Because LastPass used PBKDF2 with low iteration counts for many legacy users — and because those users had no additional entropy layer like a Secret Key — vaults became crackable offline. If you had a weak master password and no architectural defense, you were genuinely exposed. Not theoretically. Your vault was sitting on someone’s server getting hammered by a GPU cluster. That’s why 1Password’s Secret Key model and Bitwarden’s Argon2id adoption aren’t just marketing talking points — they’re direct responses to an attack vector that’s been demonstrated in the real world.

On audit history, the managers I’d actually recommend all have reasonably transparent records, though with gaps worth knowing about.

1Password does annual audits through Cure53. The 2024 review covered desktop apps, browser extensions, and backend infrastructure including cryptographic review. Published reports, no major findings in the last two cycles. Bitwarden has been audited by both Cure53 and Insight Risk Consulting — a 2022 finding around clipboard handling was patched, and the transparency throughout has been good. NordPass had a full Cure53 audit in 2023, published publicly.

Proton Pass completed its first major Cure53 audit in 2024 covering the open-source clients. The backend audit scope was narrower, which I’m still watching. The next cycle will tell us more about how seriously they’re treating this. Keeper does annual audits through Coalfire and holds SOC 2 Type II certification — broad security controls coverage, though the cryptographic implementation review isn’t as granular as Cure53’s typical work, which is a meaningful difference if you’re evaluating at a technical level. Dashlane’s last full independent audit was 2022. For a product at this market position in 2026, that’s a gap I’d raise directly with them before committing.

The question worth asking any vendor: what’s the scope of your most recent third-party audit, and where’s the published report? Both parts matter. A scope that excludes the backend tells you something. An unpublished report tells you something else.

“Best password manager” is a bit like “best car” — the answer depends entirely on what you’re actually doing with it.

Journalists, activists, or anyone with metadata exposure as a genuine concern — the two options I’d point to are Proton Pass and Bitwarden self-hosted. Proton operates under Swiss law with no advertising business model. Bitwarden self-hosted on a cheap VPS means your vault never touches their infrastructure at all. Both have open-source clients you can actually audit. In this setup, I’d also disable crash reporting in settings and pair vault access with a YubiKey 5 series or equivalent hardware key. The YubiKey setup takes about 20 minutes and most people never bother, which I find genuinely baffling given how much is at stake.

Small business owners running 5–50 person teams have different priorities: SCIM provisioning, SSO/SAML integration, role-based access, admin audit logs. The comparison narrows pretty quickly to 1Password Teams and Keeper Business. 1Password’s Admin Console has granular vault sharing policies and activity logs you can push to a SIEM. Keeper’s zero-knowledge shared folders with per-user permissions work well for client-facing teams where you need to grant and revoke access quickly. Both support enforcing MFA across all accounts, which should be non-negotiable at this level.

Family setups are a different problem again — what you actually need is easy recovery for people who aren’t particularly technical, emergency access, and low friction day to day. 1Password Families handles this well. Their Emergency Kit (a printable PDF with your Secret Key and setup instructions) is honestly the most senior-friendly recovery mechanism I’ve tested — my parents could follow it, which is not a low bar. Bitwarden Families is $40/year for 6 users, which is remarkable value. Both offer emergency access with configurable waiting periods; a 72-hour delay gives you time to deny a fraudulent request before it goes through.

Students or anyone just getting started: Bitwarden Free. Unlimited passwords, sync across devices, TOTP storage, breach monitoring — all free. The interface isn’t as polished as 1Password, but the security architecture is comparable. Upgrade to Premium for $10/year when you want health reports and advanced 2FA options. Genuinely the best $10 you’ll spend on digital security.

Remote workers where autofill reliability actually matters — this sounds like a low bar, but in practice it isn’t. I tested autofill across 50 sites. 1Password handled 47/50 correctly, including tricky embedded iframes and subdomain variations. It correctly refused to autofill on paypa1.com (a lookalike domain) while filling paypal.com without hesitation. That’s phishing resistance working quietly in the background, the way it should. Dashlane scored 44/50, but has one genuinely useful edge: the local vault cache works completely offline, which matters if you’re frequently on unreliable connections. I travel enough that this occasionally matters to me.

Your password manager is a high-value target. What you put in it is itself a security decision, and I think this gets undersold.

Cryptocurrency seed phrases — the 12 or 24-word recovery mnemonics — shouldn’t live in your vault. If the vault is compromised, the crypto wallet goes with it. Hardware wallets with offline backup on steel are the right approach. Similarly, your vault’s own recovery codes shouldn’t be inside the vault. Store those somewhere physical. And your vault master password does not belong in the vault. I’ve seen this more than once, and each time I see it I feel a little sad.

On TOTP codes: there’s an ongoing debate about whether to store them in-vault for convenience or keep them in a separate authenticator for security. My honest take is that for most accounts, in-vault TOTP is fine and dramatically improves actual compliance — people use it because it’s easy. But for your password manager login itself, your primary email, and financial accounts, use a hardware key or a separate authenticator like Ente Auth (open-source, encrypted backup). The vault shouldn’t contain its own second factor. That’s the rule I actually follow.

Migrating between managers is where things go sideways in ways people don’t anticipate. The step most people miss is TOTP seeds — many exporters silently drop them without warning. Before deleting your old manager, manually verify that at least five TOTP codes generate correctly in the new one. Export in both CSV and whatever proprietary format handles attachments. Import into a test account first, check that item counts match. Then rotate passwords for your top 20 highest-risk accounts — email, banking, work SSO at minimum. Revoke active sessions in the old manager. Secure-delete the export file (standard deletion doesn’t overwrite disk; use srm on macOS or Eraser on Windows). And wait 30 days before canceling the old subscription. I say this from experience: you will find something you forgot.

Three things worth actually doing, in rough order of how quickly you can get them done.

Today: open your manager’s settings and find the encryption section. Bitwarden users, confirm Argon2id is selected. 1Password users, find your Emergency Kit PDF and make sure it exists somewhere physical — not just on the device it’s protecting. If you can’t find what encryption your manager uses in five minutes, that’s useful information in itself.

This week: run the password health report and find your three most reused or breached passwords. Change just those three — don’t try to fix everything at once, you won’t. Also set up hardware key authentication for vault access if your manager supports it. Most do now. Almost nobody does this, and it’s probably the single highest-leverage thing you can do.

Longer term, the right setup depends on your situation. Business owner — look at 1Password Teams or Keeper Business with SSO. Privacy-focused — Bitwarden self-hosted is worth the afternoon it takes to set up, and there are good Docker guides that make it less painful than it sounds. Managing a family — build an Emergency Kit for each member and keep them somewhere fireproof, not in a drawer someone will eventually lose track of.

By late 2026, most major platforms will default to passkey prompts. Google, Apple, Microsoft, and most Fortune 500 login systems are already moving this direction. The password manager’s job is shifting from credential storage toward passkey orchestration, and the managers that figure out cross-platform passkey sync and recovery are going to pull ahead of the ones that haven’t invested there.

I’ll be honest — I’m still working out what I think about the longer-term picture. If passkeys become the default and hardware-bound keys become standard, does the traditional third-party password manager survive? Or does OS-level credential management — Apple Keychain, Windows Hello — absorb most of the use case, leaving enterprises and privacy-focused users as the remaining market? I genuinely don’t know. The FIDO Alliance’s 2025 Passkey Adoption Report has useful data if you want to form your own view without taking anyone’s word for it.

For further reading: Cure53’s published audit repository at cure53.de is worth bookmarking, and the EFF’s Surveillance Self-Defense guide is still the best practical introduction to threat modeling I’ve come across.

The question worth sitting with: do you know what happens to your passwords — and your passkeys — if you get locked out of your manager tomorrow? If you’re not sure, that’s your starting point.