What Is Ouroboros Leios? Cardano’s Path to Higher Throughput

TL;DR

• Ouroboros Leios is a proposed redesign of Cardano’s consensus that separates transaction diffusion, validation, and ordering into parallel stages, so the network uses bandwidth and CPU that Ouroboros Praos leaves idle.
• It adds two block types: endorser blocks (EBs) certify them through committee voting, and ranking blocks (RBs) anchor the chain.
• Leios preserves Praos security, holding the existing resistance to an adversary controlling up to roughly 50% of stake.
• The design under CIP-0164 is Ouroboros Linear Leios, a stepping-stone variant that keeps ledger changes minimal to reach mainnet sooner.
• Published targets range from roughly 10x to 65x current Praos capacity, framed by Input Output as simulation and testnet figures, separate from live mainnet measurement.
• Leios reached its first public SPO testnet on June 23, 2026. It requires a hard fork routed through governance, and no mainnet date is confirmed.
• Leios addresses throughput. Ouroboros Peras addresses settlement finality. They are two separate upgrades that readers often conflate.
• Stake pool operators face a modest hardware step-up (6+ CPU cores, 100 Mbps+ bandwidth) plus one new voting key. Delegation mechanics stay unchanged for ADA holders.
• Everstake has historically operated 130+ networks to date, including Cardano.

Why Cardano needs a new consensus design

Cardano’s throughput ceiling is a design consequence of Praos. The current protocol is secure and peer-reviewed, and it deliberately produces blocks in a sequential rhythm that limits how much data the base layer moves per second.

Sequential block production favors verifiability over raw speed.

Independent trackers place Cardano’s base-layer throughput near 10 transactions per second, with practical ceilings around 10 to 18 TPS for simple payments depending on transaction size.

Input Output (the company behind Cardamo and Ouroboros Leios) product manager Michael Smolenski described the current ceiling as

‘adequate for proving out the design, though insufficient for the next phase of adoption.’

The economics apply to operators too. Smolenski noted that stake pool operators need network usage closer to 50 TPS over the long term, so fee revenue can eventually offset declining block rewards.

Cardano scalability and Cardano tps are structural constraints built into the current design.

Leios is one part of a wider scaling picture. It works alongside:

• Peras for faster settlement,
• Hydra as a Layer 2 solution,
• Mithril for lightweight verification.

Understanding where Leios fits requires seeing it as a Layer 1 consensus change, operating at the base layer alongside the Layer 2 approaches that surround it.

Ouroboros Praos and the limits of one block per slot

Ouroboros Praos is the consensus protocol Cardano runs today, and it produces one block per active slot in sequence. Time is divided into slots and grouped into epochs.

For each slot, a stake-weighted lottery selects a slot leader, that leader produces a block, the block propagates and is validated, and only then does the next slot’s work proceed.

That sequential model leaves capacity idle. During the interval between producing a block and diffusing it across the network, much of the available bandwidth and CPU sits unused.

A new block every roughly 20 seconds is secure and predictable, though it caps the transaction volume the base layer can settle. Enlarging blocks or shortening slots pushes more data through the same sequential pipe, which slows the agreement process as nodes struggle to keep pace.

Leios is built to relieve this sequential-throughput constraint, which defines the Cardano consensus protocol problem Leios targets.

What Ouroboros Leios changes

Leios keeps Praos running underneath and adds an overlay that fills the idle capacity. It introduces one new block type so that carrying extra transactions, validation, and ordering no longer all happen inside a single block.

Input Output describes it as a high-throughput overlay that manages a structured flow of transactions.

In Praos, a single block does everything: it carries transactions, gets validated, and takes its place in the chain order at once.

The current specification being tested on the Musashi Dojo testnet, Ouroboros Linear Leios (CIP-0164), divides this work into two block types:

1. Endorser blocks (EBs) carry batches of additional transactions and are certified through committee voting before those transactions enter the ledger.
2. Ranking blocks (RBs) are the standard Praos blocks; they anchor certified work into the linear chain using proven Praos mechanics, and each EB is announced and coupled to the RB chain.

Because EBs are no longer produced sequentially, but produced and validated alongside the regular Praos chain, Leios fills what would otherwise be idle intervals. It raises throughput while keeping the security assumptions intended for Praos.

The fuller three-block research design (adding a separate “input block” tier for even higher concurrency) is the longer-term vision. Linear Leios deliberately eliminates input blocks and keeps EB production coupled to RBs, a non-concurrent form that keeps ledger changes minimal to reach mainnet sooner.

Praos versus Leios at a glance

Property	Ouroboros Praos	Ouroboros Leios
Block model	One block per slot, sequential	Two block types, pipelined
Diffusion and ordering	Coupled in a single block	Decoupled across tiers
Capacity use	Bandwidth and CPU often idle	Designed to fill idle capacity
Throughput framing	~10 TPS base layer	10x to 65x target (claimed)
Security model	Resists adversary up to ~50% stake	Preserves the same assumption
Status	Live on mainnet	Public testnet, no mainnet date

Types of blocks: endorser blocks, and ranking blocks

Each Leios block type has one job, and the names have changed across research iterations, so current terminology applies. The definitions below follow Input Output’s own Leios glossary and CIP-0164.

Note that earlier papers used different structures. The original Leios research paper defined three block types, including a separate “input block” (IB) tier. Linear Leios (the design specified in CIP-0164 and now running on the Musashi Dojo testnet) eliminates input blocks: endorser blocks reference transactions directly, and only two block types remain.

Endorser blocks (EBs) certify the input work. An endorser block references input data and is produced by nodes that win an endorsement lottery.

A committee of voters, typically around 500 eligible operators, checks the referenced data and votes. A quorum, often set near 60% of the committee, certifies the endorser block.

Ranking blocks (RBs) anchor everything into the chain. A ranking block is a Praos-style block that orders other blocks and forms the main chain.

A ranking block carries the aggregated certificate for a certified endorser block. If an endorser block is never certified by a ranking block, its referenced input blocks do not enter the ledger.

Concurrency, committees, and optimistic validation

Leios keeps its pipelines full by validating optimistically and certifying in committees. Producing endorser blocks alongside the regular Praos chain lets transaction diffusion and ordering overlap.

Committee certification is the trust mechanism. A defined set of stake-weighted voters casts votes on endorser blocks, and those votes aggregate into a compact certificate carried by a ranking block.

A recent CIP-0164 update shrank these certificates from roughly 8 kB to about 200 bytes, roughly 40x smaller, which eases the load on ranking blocks.

The design also handles adversarial behavior. Because Leios diffuses multiple contemporary blocks at once, it cannot assume it has seen every predecessor, so it manages equivocation, leaked keys, and burst conditions through its voting and freshest-first delivery rules.

Input Output built an adversarial testing tool named Piranha to simulate protocol attacks during development.

Security preservation is the stated boundary. The approach is constrained so that the shape of the Praos chain looks the same whether or not the Leios overlay is running.

Matching the Praos chain shape keeps the existing resistance to an adversary controlling up to roughly 50% of stake.

The throughput claims, and how to read them

Every Leios throughput figure is a claim attributed to a source, and each should be read as simulation, target, or measurement. Input Output’s own materials present these numbers as design targets validated in simulation and early testing.

Understanding how execution models shape blockchain scalability helps frame why these numbers vary so widely.

The published targets cluster around a wide range:

• Input Output’s roadmap lists transaction data throughput targets of 140 to 300+ TxkB/s, described as 30x to 65x current Praos levels.
• The same roadmap cites peak throughput of 1,500+ TPS under what it calls realistic conditions.
• The treasury proposal frames a phased mainnet increase scaling from 2x to 30x current capacity, with fuller capacity demonstrated on testnet first.
• A July 2025 simulation stress test reported 1,000 TPS with non-Plutus transactions in a controlled setting.

Figures like 1,000 TPS or 65x come from simulations, stress tests, and testnet runs, separate from sustained mainnet operation.

The initial mainnet release has been described by Input Output as closer to a 50x improvement over today’s baseline, with further increases unlocked gradually through parameter changes.

Leios and Peras: two upgrades, two problems

Leios and Ouroboros Peras are two separate upgrades that solve different problems. Leios raises throughput. Peras shortens the time before a transaction is considered final.

The two upgrades can operate together, and function independently .

Peras targets settlement time. Under Praos, Cardano uses Nakamoto-style probabilistic settlement, where confidence that a block will not be rolled back grows as more blocks build on top.

Theoretical high-assurance settlement can require a long wait. Peras adds a committee voting layer that certifies blocks, cutting high-confidence settlement to around 2 minutes.

The two upgrades share infrastructure while staying distinct:

Dimension	Ouroboros Leios	Ouroboros Peras
Problem solved	Transaction throughput	Settlement finality
Mechanism	Parallel endorser, ranking blocks	Committee voting to boost blocks
Headline figure	10x to 65x capacity (claimed)	~2-minute high-confidence settlement
Layer	Layer 1 consensus change	Layer 1 consensus change

Both operate at Layer 1, changing consensus at the base layer. Hydra operates at Layer 2, so describing Leios as a Layer 2 alternative is inaccurate: Leios changes the base layer itself.

Readers following how another major network is replacing its consensus protocol with Alpenglow replacing Tower BFT will recognize the same pattern of a base-layer consensus redesign.

Where Leios stands now

Leios is a research and engineering programme, and no mainnet date is confirmed. The design is specified in CIP-0164, titled Ouroboros Linear Leios, published in August 2025 and since merged.

The pivot to the Linear variant traded some long-term throughput headroom for a shorter path to deployment by avoiding ledger changes that would require ecosystem-wide adaptation.

During the budget cycle ending June 2026, CIP-0164 was merged, initial throughput was demonstrated on an alpha implementation, and Leios was deployed to a dedicated public network.

The first public SPO testnet went live on June 23, 2026, the first time Leios ran on a public network outside simulation. Community treasury funding of ₳27,714,342 was approved with roughly 84% support.

The road to mainnet runs through defined stages. The 2026/27 cycle aims to mature Leios from prototype to a mainnet-ready release candidate, progressing through Software Readiness Levels 5 to 8, with adversarial testing, conformance checks against the formal specification, and testnet hard forks along the way.

Because a hard fork routes through governance, activation also depends on Cardano’s on-chain governance process and the Cardano roadmap 2026.

A hard fork is required, and that is the final gate. Leios changes consensus, so mainnet activation needs a hard fork enabled through Cardano’s governance system.

Input Output has publicly targeted 2026 for mainnet, and the schedule is milestone-gated with no fixed date confirmed.

What changes for stake pool operators and delegators

Leios raises node requirements modestly for stake pool operators, and the changes are designed to stay within reach of typical operators. Input Output’s published guidance recommends:

• 6+ CPU cores,
• 100 Mbps+ bandwidth,
• SSD storage.

It is a modest upgrade over current requirements. The stated goal is to avoid the node-accessibility trade-offs seen on some higher-throughput chains.

Operators also take on new protocol duties. To participate in Leios voting and block production, stake pool operators must register one additional cryptographic key for the voting scheme.

Producing input blocks, participating in endorsement committees, and handling the new block types add operational responsibility beyond today’s Praos workflow.

Fee structure is part of the design conversation. Higher sustained throughput is meant to support an eventual move toward fee-driven operator economics, since Input Output has framed rising network usage as necessary for operators to cover costs as block rewards decline over time.

For delegators, the core experience stays the same. Delegating ADA to a stake pool works the same way under Leios as under Praos.

Stake stays non-custodial, delegation mechanics are unchanged, and delegators take no action to benefit from higher base-layer capacity. This preserves the existing Cardano proof of stake consensus experience for the average participant.

Everstake has historically operated validator infrastructure across 130+ networks to date, including Cardano, and follows protocol upgrades like Leios so operators and delegators can assess what changes and what does not.

FAQ

What is Ouroboros Leios?

Ouroboros Leios is a proposed redesign of Cardano’s consensus protocol that separates transaction diffusion, validation, and ordering into parallel stages. It adds endorser, and ranking blocks so the network uses bandwidth and CPU that Praos leaves idle, raising throughput while preserving existing security assumptions.

How is Leios different from Praos?

Praos produces one block per slot in sequence, coupling transaction data, validation, and ordering into a single block. Leios decouples those jobs across three block types that run concurrently, filling idle capacity between blocks.

What is the difference between Leios and Peras?

Leios increases transaction throughput. Ouroboros Peras shortens settlement finality to around 2 minutes using committee voting. They are two distinct Layer 1 upgrades that can work together while solving different problems.

Will Leios require a hard fork?

Yes. Leios changes consensus, so mainnet activation with Leios requires a hard fork enabled through Cardano’s on-chain governance process.

When will Leios reach mainnet?

No mainnet date is confirmed for Leios. It was previously targeted 2026, and the first public SPO testnet went live on June 23, 2026, with activation gated by testing, audits, and governance approval.

How much will Leios increase Cardano throughput?

Published Leios targets range from about 10x to 65x current Praos capacity, with roadmap figures citing 140 to 300+ TxkB/s and peaks of 1,500+ TPS under simulated conditions.

Does Leios replace Layer 2 solutions like Hydra?

No. Leios is a Layer 1 consensus change to Cardano’s base layer, and Hydra is a Layer 2 scaling solution. They operate at different layers and are distinct tools.

What do stake pool operators need for Leios?

Input Output recommends 6+ CPU cores, 100 Mbps+ bandwidth, and SSD storage for Leios, plus one additional cryptographic key registered for Leios voting. 

Disclaimer:
Everstake is a software platform that provides infrastructure tools and resources for users but does not offer investment advice or investment opportunities, manage funds, facilitate collective investment schemes, provide financial services or take custody of, or otherwise hold or manage, customer assets. Everstake does not conduct any independent diligence on or substantive review of any blockchain asset, digital currency, cryptocurrency or associated funds. Everstake’s provision of technology services allowing a user to stake digital assets is not an endorsement or a recommendation of any digital assets by it. Users are fully and solely responsible for evaluating whether to stake digital assets.