Falcon-1024 Accounts Move From Bolt-On to Built-In
Algorand's relationship with post-quantum cryptography did not start this month. State Proofs, the mechanism that anchors Algorand's entire transaction history back to genesis, have used FALCON signatures since 2022, years ahead of most chains even discussing quantum migration. The network processed its first quantum-resilient mainnet transaction in 2025. In June 2026, the Algorand Foundation published a formal post-quantum roadmap targeting broad quantum resilience by 2027. Native Falcon-1024 accounts in v5.0.0 are the first concrete deliverable off that roadmap, not a first step taken from scratch.
What actually changes for a user is narrower than the headlines suggest. Previously, quantum-resistant protection required custom logic layered on top of a standard account, an opt-in workaround rather than a protocol primitive. With v5.0.0, an account can be created with Falcon-1024 signatures directly, no extra scaffolding required. These addresses use a separate scheme from classical keys, so a traditional key can never accidentally satisfy a post-quantum account's signature check. That separation matters more than it sounds: it closes a class of downgrade attack where an attacker tries to get a quantum-resistant account to accept a classically-signed transaction instead.
The Algorand Foundation was explicit that this is not a finish line. Other parts of the protocol, consensus signatures and verifiable random functions among them, are still migrating over time, and the Foundation's own messaging stopped short of claiming full future-proofing. The "harvest now, decrypt later" threat model, where an adversary records encrypted data today to break it once sufficiently powerful quantum computers exist, is a multi-year migration problem for every chain, not a box any single release checks off.
Why Fees Had to Change First
Here is the part easy to miss if you read the account upgrade and the fee upgrade as two separate stories: quantum-resistant signatures are cheaper to verify computationally than standard signatures, but they take up meaningfully more space in a block. Under Algorand's old flat-fee model, every transaction paid the same base rate regardless of what it actually cost the network to process and store. That made a Falcon-1024 transaction, larger but not more expensive to compute, effectively subsidized by everyone sending ordinary payments, or alternatively, it made pricing post-quantum accounts fairly impossible without either overcharging simple transfers or undercharging the transactions that actually strain block space.
Version 5.0.0 replaces the flat rate with a model that charges based on the resources a transaction consumes: larger data payloads and heavier computation cost more, and those fees flow into a shared pool that pays node operators. The Foundation's framing ties this directly to long-term network sustainability, part of the thinking behind an upcoming paper the Foundation calls King Safety. The practical promise is that everyday activity, simple payments and basic transfers, stays exactly as cheap as it has always been. Algorand's transaction fees have long been a selling point against congested L1s, and the Foundation's messaging is explicit that this upgrade is not a backdoor fee hike for ordinary users. It is a repricing of the specific things that cost more to process.
| Component | Before v5.0.0 | After v5.0.0 |
|---|---|---|
| Post-quantum accounts | Custom logic layered on a standard account | Native Falcon-1024 address type |
| Transaction fees | Flat rate regardless of resource use | Priced by data size and computation |
| Smart contract size limit | Baseline AVM ceiling | Doubled |
| Box storage visibility | Isolated per application | Optional cross-app read/collaboration |
| Activation threshold reached | August 15, 2026 (90% node support) | |
| Mainnet activation | After ~208,000 round (~7 day) cooldown | |
AVM v13: More Room, and Apps That Can Finally Talk to Each Other
The same repricing logic unlocked a set of developer-facing changes that the Foundation says simply were not practical to offer before, because there was no fair way to charge for them. AVM v13 doubles the maximum smart contract size, letting existing applications grow in place rather than forcing developers into painful rebuilds once they hit the old ceiling. That matters more for mature protocols with years of accumulated logic than for new deployments, which rarely bump into size limits on day one.
More interesting architecturally is the change to box storage. Box storage is an application's private data store, and until now it was invisible to every other app on the network, unlike global and local state, which have always been externally readable. Developers can now opt into letting trusted applications read from, or collaborate around, their box storage. That is a meaningful primitive for anything building composable DeFi or multi-contract systems where apps need to coordinate state without routing everything through a shared intermediary contract.
The release also adds a poseidon2 hashing opcode, tuned for zero-knowledge proof systems, positioning Algorand for the growing set of applications that want to prove something is true without revealing the underlying data. And it introduces an early network-load signal, described by the Foundation as a preview rather than a functioning mechanism yet, that lays groundwork for future congestion-aware systems without doing anything active in this release.
What the Cooldown Period Actually Means
Passing the 90% support threshold is a governance milestone, not an activation event, and conflating the two is the most common mistake in coverage of consensus upgrades generally. The 208,000-round cooldown, roughly a week, exists specifically to give the remaining node operators time to upgrade before the new rules become binding for everyone. Community reports put thousands of node runners upgrading within days of the threshold being hit, which is a healthy signal for how contentious (or uncontentious) this upgrade is within the operator base, but it does not change the mechanical fact that none of these features, not the Falcon-1024 accounts, not the new fee model, not AVM v13, are live on mainnet as of this writing.
That distinction is worth holding onto for a specific reason: it is also the largest protocol upgrade Algorand has shipped since staking rewards went live in January 2025, which itself took time to show its effects on network participation and token economics. Users and developers evaluating what changes for them should wait for the actual activation date rather than building around a feature set that is still in cooldown.
Key Takeaway
The fee model is the upgrade, not a side effect of it: Native post-quantum accounts and the expanded AVM toolkit exist because usage-based fees made it possible to price the extra block space and computation they require. Read the three pieces of this release as one architectural decision, not three separate headlines.
Ordinary users see little immediate change: The Foundation was explicit that everyday payments and basic transfers stay exactly as affordable as before. The repricing targets larger data payloads and heavier computation, not standard activity.
90% support is not activation: v5.0.0 cleared its threshold on August 15 and enters a roughly seven-day cooldown before going live on mainnet. None of the new features are active yet.
This is a milestone in a multi-year migration, not a finish line: Falcon-1024 accounts are the first deliverable from a post-quantum roadmap targeting 2027, and the Foundation itself says other protocol components are still migrating. Full quantum resilience is not achieved with this release.
Developers get room they did not have before: A doubled contract size limit and optional cross-app box storage access are the kind of primitives that matter most to mature, complex applications rather than simple new deployments.