Cardano Leios Enters Measurable Engineering Phase as Prototype, CLI Work and Benchmarks Advance

Cardano’s Leios scaling roadmap is moving from research design into measurable engineering. During the April 2026 Leios Monthly Review, held on April 29, IOG developers and contributors from the wider Cardano community presented progress across the prototype, transaction validation benchmarks, network testing, Cardano CLI support and committee selection design.

The message from the review was not that Leios is finished, or that Cardano mainnet is already operating at a new throughput level. It was more specific and more important. The components needed to test Leios as a real scaling architecture are beginning to line up, from Extended Blocks and vote diffusion to ledger benchmarks, mempool behavior, BLS keys and end to end test infrastructure.

For Cardano, this matters because Leios is not just another performance proposal. It is one of the ecosystem’s most important attempts to increase throughput while preserving the security and decentralization assumptions that define the network. The April review showed that this work is now being tested through concrete engineering questions rather than broad scaling promises.

Cardano Leios Prototype Shows Early Throughput Progress

The review opened with a look at the Leios prototype and how Extended Blocks, known as EBs, are being sequenced alongside the existing Cardano Praos system. In the live development setup, the team showed the network operating under a load of around 20 transaction kilobytes per second, described in that specific scenario as roughly 100 transactions per second on smaller transactions.

That number should not be treated as a final Leios performance claim. The team is still working toward a higher controlled environment target of 200 transaction kilobytes per second. In a theoretical example using very small 200 byte transactions, that could correspond to around 1000 transactions per second, but the review did not present that as an achieved mainnet result.

The more important signal was how the prototype behaved. The visual demo showed Praos blocks and Leios Extended Blocks appearing in the same environment, with the mempool being depleted while confirmed throughput followed the load placed on the system. That gives the development team a clearer way to observe whether Leios is actually moving transaction data through the system as intended.

The review also showed progress on vote diffusion visualization. Developers can now inspect how votes move across the network, what is in flight between nodes and how specific EB hashes are being voted on by participating identities. This matters because Leios cannot be judged only by how much data it can introduce. It also has to prove that the network can propagate, vote on and certify that data reliably.

This is the hard part of Cardano scaling. A higher throughput target is easy to write down. A decentralized protocol that coordinates larger data flow across real nodes, bandwidth limits, latency differences and adversarial conditions is a different challenge. The April review showed that Leios is now being tested against that second problem.

Cardano Benchmarks Focus on Validation, Mempool Efficiency and Network Load

A major part of the review focused on benchmark work. The ledger transaction validation benchmarks are designed to measure the cost of different validation strategies using an actual Cardano ledger implementation. These tests do not measure the full network. They isolate specific parts of the system so the team can understand where optimization may be necessary.

The results showed meaningful differences between validation approaches. For small value transactions, the review discussed an approximate five times speedup when moving away from full validation in the tested scenario. For script transactions, the difference was much larger, with an approximate fifty times speedup mentioned in relation to the benchmark results.

Those figures do not mean every real workload will behave in the same way. They do show why the benchmark work is important. If Leios is expected to increase the amount of transaction data moving through Cardano, the network also needs to understand how validation costs scale across transaction types, block sizes and hardware conditions.

The review also included transaction submission v2 testing. The current v1 model used on mainnet was described as simple but wasteful because multiple peers can download the same transactions after receiving the same announcements. That creates duplicate data movement and lowers network efficiency.

The newer v2 on decision approach showed much stronger results in the presented tests. In one run, it completed the transfer in about 1.6 seconds with roughly 78 percent efficiency, compared with v1 results where only around 9 percent of the data carried was useful. That improvement is relevant for Leios, but it could also improve Cardano more broadly because better transaction submission reduces network waste before any full Leios activation.

The team is also preparing larger network testing through the PNT cluster, using more realistic node topology, regional distribution and latency profiles. The goal is to understand how much transaction data the current Cardano node and networking stack can handle before the Leios consensus layer is fully introduced.

This is where the review became more than a prototype update. The work is now looking at the parts of scaling that usually determine whether a protocol can leave the lab, mempool replication, duplicate downloads, transaction submission, trace messages, network latency and the real cost of moving more data through a decentralized system.

Leios Advances Cardano CLI Support and Committee Design

The April review also showed that Leios preparation is reaching the tooling layer. The team demonstrated Cardano CLI support for BLS keys, which will be required for participation in the Leios consensus protocol.

Stake pool operators will need to generate a BLS key and a proof of possession in addition to their usual keys. The proof of possession is important because it helps protect against rogue key attacks. According to the review, the relevant commands are already available in the latest Cardano CLI release under the Dijkstra era.

The expected process is that, after the relevant hard fork, SPOs will register the new BLS credentials through a stake pool registration certificate and include the BLS signing key in the node startup configuration. That is a practical sign of progress. Leios is no longer only a research discussion. It is beginning to touch the operational workflow that Cardano infrastructure providers will eventually need to follow.

Another important update came from the committee selection discussion. The team is considering a change from the current SIP164 proposal, which uses a more complex weighted model with local sortition fallback. The alternative being discussed would be simpler, easier to implement and more efficient to verify.

The reason is practical. Under the simpler model, votes can be smaller, verification can be faster and certificates can be significantly reduced in size. The review discussed a comparison where certificate size could move from around 8 kilobytes to around 200 bytes, depending on the scheme and parameters.

That detail matters because scaling is not only about increasing transaction capacity. If higher throughput creates heavy voting overhead, large certificates or expensive verification paths, the network can gain performance in one area while losing efficiency in another. By simplifying committee selection before production pressure arrives, the Leios team is trying to remove complexity at the protocol level rather than manage it later through operational workarounds.

The review also highlighted end to end testing readiness. Cardano already has more than 2000 system level test scenarios covering areas such as Cardano node, Cardano CLI, db sync, governance, delegation, rewards and rollback behavior. The plan is to use this testing base for Leios, with local testnet variants that more closely resemble mainnet parameters and with transaction generators creating sustained load.

The next phase is therefore becoming clearer. Leios now has to move from prototype behavior and component benchmarks into broader testnet conditions, where throughput, vote diffusion, committee selection, CLI readiness, network propagation and SPO operations can be evaluated together. That is the real checkpoint. Not a headline TPS number, but whether the full system can behave predictably when the pieces are connected.

For Cardano, the April Leios review marks progress because it shows the scaling roadmap becoming operational. The protocol is being tested through the same layers that will decide its real value, the ledger, the mempool, the network stack, the committee model, the CLI and the operators who will run it. If those layers continue to converge, Leios will not be just a scaling concept for Cardano. It will become a testable path toward higher capacity without abandoning the engineering discipline that made the network different in the first place.