Cardano Leios targets 3–5x capacity through staged Dijkstra rollout

Input Output has outlined a staged rollout for Cardano’s Leios upgrade, targeting an initial three to five times today’s transaction capacity before later governance votes could raise it to 10–15 times.

Input Output said in its Oct. 20 update that Leios will arrive with the Dijkstra hard fork, followed by stake pool registration, live network load exercises, and at least two protocol parameter updates.

The developer’s plan separates the network upgrade from the steps needed to activate Leios and permit more traffic. Pools must first register new voting keys, with the current estimate requiring roughly 90–95% of delegated stake before Leios can operate.

“The hard fork makes the protocol available,” Input Output wrote. The update gives no fixed Dijkstra activation date. Its proposed capacity gains remain subject to mainnet results and community approval.

Cardano Leios needs pool registration after Dijkstra

Under Input Output’s rollout plan, stake pool operators will generate BLS voting keys after the hard fork and register them through updated pool certificates. Each registration becomes active at the next epoch boundary.

BLS signatures allow hundreds of pool votes to be combined into a compact certificate. Input Output says Leios needs votes representing at least 75% of all active stake to certify an endorser block, which carries references to transactions awaiting processing.

Registration must sit comfortably above that threshold to allow for pools that are offline or slow to respond. The developer estimates participation from roughly the top 500–900 pools, although the exact activation requirement depends on parameters still being finalized.

Pools without a registered BLS key can continue producing Praos blocks, according to the update, but cannot vote on Leios endorser blocks. Key registration requires no separate vote from delegated representatives or the Constitutional Committee.

The preparation follows Cardano’s July 18 van Rossem hard fork, which moved mainnet to Protocol Version 11, as crypto.news previously reported. Intersect confirmed that upgrade after testing and approval from the network’s governance bodies.

Capacity changes require separate governance approval

Before Dijkstra, Input Output plans a constitutional update covering the new protocol parameters and the limits within which governance can change them. Its roadmap assigns preparation of those guardrails to the Intersect Parameters Committee, with the Plutus team updating the guardrails script.

The developer lists different approval thresholds for each stage. The constitutional amendment needs support from delegated representatives representing 75% of stake and two-thirds of the Constitutional Committee.

For the hard fork, the stated requirements are 60% of DRep stake, 51% of stake pool operator stake, and two-thirds of the committee. Later throughput changes require 67% of DRep stake, 51% of SPO stake, and two-thirds of committee members.

Once Leios is operating, Input Output plans controlled bursts of mainnet traffic to measure block delivery, voting speed, certificate reliability and the processing room left on nodes. Each proposed capacity step should include measurements from the preceding settings.

The initial target uses today’s Praos ceiling of roughly 4.5 kilobytes of transactions per second as its baseline. Multiplying that figure by the announced ranges gives approximately 13.5–22.5 TxkB/s initially and 45–67.5 TxkB/s at the later target.

These figures measure transaction data capacity, not a fixed number of transactions per second. The number of transactions that fit depends on their size.

Input Output says governance can change the Leios limits without another hard fork, provided the proposals remain within the approved constitutional guardrails.

Earlier testing reached roughly six times the Praos ceiling

In an Aug. 20 engineering account, the Leios team recorded a MusashiNet peak of 26.8 TxkB/s against a Praos mainnet ceiling of 4.51 TxkB/s. The figures produce a ratio of approximately 5.94 times.

The team said the 41-day Earth phase involved more than 127,000 blocks and 63 registered pools. Traffic came from synthetic load generation and participating operators, so the results describe testnet performance rather than customer demand.

Its account documented implementation problems, including memory leaks, chain forks, stalled synchronization, and a network halt caused by handling an invalid certificate as a fatal error. The team said it corrected the faults.

For mainnet, Input Output’s Oct. 20 plan starts with conservative settings because the live network includes approximately 2,700 registered pools using different machines, connections and configurations.

Leios retains Praos ranking blocks and adds endorser blocks when pending traffic exceeds what a ranking block can carry, the developer explained. Transactions processed through endorser blocks are expected to reach the ledger in roughly 45–60 seconds.

Dijkstra preparation has proceeded through separate software releases. September’s node 11.1.1 release addressed operational issues ahead of later versions intended to introduce the new era’s features.

Intersect’s earlier two-phase Dijkstra roadmap placed Linear Leios in the first phase and Peras in a separate upgrade targeted for the second quarter of 2027. Its schedule described delivery dates as estimates subject to testing and governance.

App leaderboard covers projects with tracking enabled

In a separate Oct. 20 X post, Cardano said 21 of the 103 apps listed on its website had enabled transaction tracking. Projects appear on the leaderboard once tracking is switched on, and participation is open to all projects building on Cardano.

The figures put tracking participation at approximately 20.4% of listed apps. Cardano’s statement describes which projects contribute to the leaderboard; it does not establish transaction totals for the remaining 82 apps.

On network economics, Input Output said more than 85% of epoch rewards currently come from Cardano’s Reserve. Its update cited a CIP-164 estimate that fee income would require sustained traffic of roughly 50–70 TxkB/s to reach break-even.

“Leios creates a route to that level of activity,” the developer wrote. “It does not create demand by itself.”

Dijkstra’s other planned features include nested transactions, account enhancements, guard scripts, Plutus V4, freely withdrawable staking rewards and a pledge-leverage parameter. According to Input Output, those changes arrive with the hard fork and do not depend on Leios reaching its later capacity targets.