Treasury funding enables the Cardano community to invest directly in the network’s continued development. In Q2 2026 (April to June), Input Output (IO) delivered engineering milestones across core stack and node reliability, developer tooling, smart contracts, and Cardano’s scaling roadmap. Two initiatives, Acropolis and Tiered Pricing, concluded as announced in April, with the unspent funding (4.1 million ada) returned to the treasury.
Key deliverables include:
- Leios testnet went public. The Musashi Dojo testnet launched on June 23, the first time Ouroboros Leios has run on a live test network.
- Reliable core stack. Cardano node versions 10.7.1 and 11.0.1 maintained steady performance; LSM and the KES agent completed beta support, with LSM reducing memory usage by roughly 65%; and the outdated tracing system was fully retired.
- Mithril and Hydra both advanced. Mithril's zero-knowledge proof work moved from prototype to testnet, and Hydra shipped two node releases carrying partial fanout and a simplified head-opening protocol.
- Builders got safer, cheaper tools. Plutus introduced a native `Value` type, and Cardano High Assurance delivered a formalized ledger, the specification of the Universal Annotation Language, a working property-based testing library, and a completed static-analysis rulebook.
- Governance-facing proposals reached the community. IO published and demonstrated a revised stake pool incentive design to SPOs and DReps, and DReps ratified IOR's Cardano Vision 2026 proposal with 74.96% support.
Leios testnet goes live with Musashi Dojo
On June 23, IO launched the Musashi Dojo public testnet for Ouroboros Leios, in collaboration with the Cardano Foundation, Intersect, Sundae Labs, Tweag, and Blink Labs. It's the first time Leios has run on a public test network instead of a local simulation, building on the working prototype demonstrated in Q1. IO operates the network while independent stake pool operators (SPOs) run Leios-enabled block producers, and builders test DApps, wallets, and infrastructure ahead of the eventual move to mainnet.
The program runs through five phases named after chapters of Musashi's Book of Five Rings: Earth, Water, Fire, Wind, and Void, moving from basic protocol validation through adversarial testing to final mainnet preparation. IO is targeting an initial five to 20 times increase in Cardano's base-layer throughput capacity, with a mainnet hard fork planned for later in 2026.
The core stack stayed reliable, secure, and fair
Cardano node v.10.7.1 shipped in April with steady gains in block production and diffusion, alongside a substantial drop in CPU usage. Node v.11.0.1 followed under protocol version 11 with no regressions. IO fully retired the legacy tracing system, cutting roughly 11,000 lines of outdated code from cardano-node.
LSM and the KES agent both completed their beta support periods this quarter: LSM cut resident memory use by around 65% remained stable, and the KES agent, which separates key management from the node process, is now ready for use on mainnet block producers. Ledger-HD's consensus redesign is complete and reviewable, with its merge deliberately timed to avoid colliding with Leios.
IO also published a full design-evaluation report on Cardano's stake-pool reward mechanism, identifying nine improvements to the current pledge-incentive formula, and demonstrated it to the community and DReps. The proposed update is intended to move into a CIP later this year. Work on nested transactions (Babel fees) continued, with the Plutus context milestone now scheduled for Q3 2026, alongside the rest of the Dijkstra-era feature set.
Plutus and Cardano High Assurance matured
Plutus introduced `Value` as a native built-in type, removing the need for language authors across the ecosystem, including Aiken, Plutarch, and Helios, to maintain their own multi-asset libraries; it will activate at the Van Rossem hard fork. A faster Plinth compiler and an updated UPLC command-line tool followed, cutting execution costs by over 10% on average, and up to nearly 50% on real mainnet scripts.
Cardano High Assurance completed the formalization of the Cardano ledger and specified a Universal Annotation Language for expressing smart contract correctness properties. A property-based testing library shipped, generating adversarial test scenarios automatically via QuickCheck, with a CLI workflow and a VS Code extension to come in Q3. The static analyzer Plu-Stan's rulebook is finished, backed by two independently compiled sets of audit-informed rules, and now ships with release binaries and a VS Code extension. The Transaction Monitoring System's operator-facing dashboard also shipped this quarter, with its AI-based detection backend and alerting layer targeted for Q3.
Mithril and Hydra advanced
Mithril's existing certificate system remains in production, and its zero-knowledge proof work moved from an April prototype to testnet by late June, with a recursive SNARK system compressing entire certificate chains and new certification support for individual blocks and transactions. On June 26, responsibility for Mithril's specialist development moved to Teragone Solutions, the team that's worked closely with the protocol since its earliest releases.
Hydra shipped two node releases: 2.1.0 in May, with improved deposit security and a new SQLite-backed event store, and 2.2.0 in June, carrying partial fanout (removing the cap on how large a head's UTXO set can grow before settlement), a simplified head-opening protocol, a revamped terminal interface, and a public, reproducible benchmark suite.
Read the report
The full Q2 2026 delivery report goes into greater depth on each of the items above.
Governance means making decisions on real evidence. This is the evidence.
