Kaspa's roadmap is organized around four pillars: throughput scaling (Crescendo), programmability (Toccata), adaptive consensus (DAGKnight), and native ZK-verified smart contracts (vProgs). This is an honest, data-driven snapshot of where each piece actually stands as of August 24, 2026 — what is live, what is still in research, and what has slipped.
| Milestone | Originally Scheduled | Actual Status |
|---|---|---|
| Crescendo (1 → 10 BPS) | May 2025 | ✅ Complete — activated May 5, 2025 |
| Rusty Kaspa v1.1.0 (sync/perf) | March 2026 | ✅ Complete — released March 4, 2026 |
| Toccata Hard Fork | 2026 H2 | ✅ Complete — activated June 30, 2026 (DAA 474,165,565) |
| DAGKnight adaptive consensus | End of Q3 2026 | 🟡 In research — KIP-2 still Proposed; target slipped |
| 32 BPS intermediate throughput | 2027 | 🔬 Research / planned |
| 100 BPS (blocks per second) | 2027 | 🔬 Research / planned |
| vProgs (native ZK smart contracts) | 2027 | 🔬 Research stage (PoC demo done Feb 2026) |
The Crescendo hard fork activated at DAA score 110,165,000 on May 5, 2025, taking Kaspa from 1 block per second to 10 blocks per second. It was shipped as Mainnet Crescendo Release v1.0.0. This was the first major throughput milestone and showed that GHOSTDAG consensus could sustain far higher block rates than a classical single chain.
Published March 4, 2026, the rusty-kaspa v1.1.0 release materially improved node and integrator performance. The headline changes were the new GetVirtualChainFromBlockV2 RPC, a reworked initial block download / catchup flow, a pruning-proof algorithm refactor, and compressed level parents plus on-the-fly higher-level relations — which cut storage pressure and sped up header processing (contributors reported up to ~3x faster header-stage IBD on some machines). It is the infrastructure layer that makes higher BPS and smoother syncs feasible.
The Toccata hard fork activated on mainnet at DAA score 474,165,565 on June 30, 2026, shipped as Mainnet Toccata Release v2.0.0 (with v2.0.1 following as a maintenance release). Crucially, Toccata keeps Kaspa UTXO-native — it does not turn Kaspa into an EVM or an account-chain smart contract system. Instead it makes UTXOs expressive enough to carry real application state.
Key capabilities now live on L1:
See our Toccata Hard Fork Guide for full details.
A hard fork being "live" is only the first step. The meaningful question is whether the new machinery is actually being used. In the weeks after activation, Kaspa's Layer 1 did see real — if early — covenant activity, which is the strongest signal that the surface has traction. This is worth reading carefully: the activity is real, but the scale is still small.
Context matters: the total live covenant value (~1.32 million KAS) is roughly $35,000 at the ~$0.0267 current price, and the 67 markets sum to about 515,418 KAS of exit value. This is a working token-launch venue, not deep market infrastructure. Lending is still absent. The balance-sheet takeaway is that the protocol surface is being genuinely exercised, but it remains early.
DAGKnight (proposed in KIP-2) is the next major consensus upgrade and a parameterless generalization of Nakamoto consensus. It is designed to tolerate up to ~50% Byzantine corruption (the paper's formal bound is up to 50% of network hashrate), it has no hardcoded in-protocol latency parameter (it adapts to the real network latency / "k" instead), and it targets sub-second finality with net-split safety. It is a prerequisite for both 32 BPS / 100 BPS throughput and safe vProgs execution, which need deterministic ordering and strong finality guarantees.
It is important to be accurate here, because this is the single most commonly over-stated item on the Kaspa roadmap. As of late August 2026:
Status: Proposed — it has not become an accepted consensus KIP.dagknight branch has recent commits (e.g. promoting Universal Merge Consensus work in mid-August 2026), so development is ongoing — it's just not shipping.In other words: this is a research-and-prototype milestone with real engineering momentum, but no committed activation. Anyone shopping for "instant finality today" on Kaspa should treat that as a forward-looking claim, not a shipped feature.
The longer-term targets build on DAGKnight. Getting from 10 BPS to 100 BPS is not a single jump — it is expected to pass through an intermediate step toward 32 BPS first. The 2027 objective of a ~10ms block time (100 blocks per second) would likely require further DAG-level algorithmic adjustments (how miners reference DAG tips) plus additional node performance work. At 100 BPS Kaspa would process thousands of transactions per second, broadly competitive with major payment networks.
vProgs (verifiable programs) are Kaspa's proposed native L1 smart contract system. A proof-of-concept was demonstrated on February 19, 2026 by Maxim Biryukov, showing a deposit-transfer-withdraw cycle for a ZK covenant rollup. However, vProgs have not reached testnet and remain a research-stage 2027 target. Key design ideas:
An important nuance: the L1 covenant toolchain that exists today already lets separate compiled covenant apps compose atomically in all-or-nothing transactions (so-called Inter-Covenant Communication). What vProgs still add on top is shared mutable state — the case where many users change one app's state at once. That is a meaningful distinction for gauging how ready "programmability" really is. See our vProgs guide for more.
Parallel to the L1 roadmap, Kasplex launched the first zkEVM Layer 2 built on Kaspa, going live during Malaysia Blockchain Week in August 2025. Kasplex is backed by the Kaspa Ecosystem Foundation and settles ZK proofs on Kaspa L1, offering EVM compatibility so existing Ethereum dApps can deploy. It uses bridged KAS as its gas token. Kasplex remains the clearest current path to EVM-style DeFi on Kaspa while vProgs target native L1 programmability. See our Kasplex L2 Guide for details.
Two market-facing events in August 2026 deserve a grounded framing.
On August 5, 2026 the network passed DAA score 504,909,000, triggering a scheduled emission step down to 23.12465141 KAS per second. The per-block reward dropped from 2.44997148 to 2.31246515 KAS. This is a routine component of Kaspa's emission schedule — the network steps down by a fixed factor (~5.61%) roughly every 2,629,800 seconds, driven by DAA score rather than by the calendar. It is not a dramatic "halving" event, and treating it as one would be misleading. The next step is 21.82676446 KAS/sec at DAA 531,207,000, roughly early September 2026.
On August 19, 2026, Coinbase announced it would delist nine perpetual futures contracts — including the KAS perpetual — effective September 3, 2026 at 13:00 UTC. It also listed POPCAT, BRETT, and several others in the same batch. The exchange framed it as a routine product review citing low liquidity and compliance risk. Critically:
Holders of the KAS perpetual were advised to close positions before September 3 to avoid automatic settlement or liquidation.
The most balanced way to read the 2026 roadmap is to separate protocol capability from ecosystem adoption — the two are not the same, and it is easy to confuse them.
Bottom line: Kaspa's engineering is real and the capabilities are shipping. But the value of that engineering depends on apps actually using it — and that layer is still being built. Treat the roadmap as a direction with real risk, not as a delivery schedule.