An Open-Source Alternative to Hyperliquid: The Case for Solidus
Hyperliquid proved on-chain perps work at CEX latency. Solidus is the open-source, permissionless, multi-algorithm alternative with trustless custody.
Status note
Solidus is in active development with mainnet pending. The performance claims in this post — sub-second block times, ~100ms finality — are measured on Solidus's test network, not aspirational design targets. The 16-chain architecture, DAGKNIGHT integration, POEM multi-algorithm mining, and ForumCore perps engine are implemented and under test. Hyperliquid's volume and open-interest figures cited below come from public production data (Artemis, The Block) and reflect real mainnet activity. Treat this post as Solidus-tested vs. Hyperliquid-shipped — both sides are real, one is further along the mainnet timeline.
The elephant in the room
Hyperliquid is the most successful on-chain perpetuals exchange of the current cycle. As of April 2026 it handles roughly $150-200 billion in monthly trading volume and has captured the majority of perpetual-DEX open interest (volume share is ~44% post-Aster disruption; OI share has run as high as 70%+). It targets ~200K orders/sec on its own Layer 1 — repeatedly demonstrating that a pure on-chain order book is technically viable at CEX-comparable performance. This is a significant accomplishment.
It is also a chain where:
- The validator node binary is available (github.com/hyperliquid-dex/node) and used by third-party RPC providers, but the HyperCore matching engine is closed-source
- Validator participation is effectively permissioned via a KYC/KYB Foundation Delegation Program — 21 validators as of early 2026 — with formal decentralization weaker than permissionless L1s
- Users must trust the operator-controlled matching engine not to front-run, censor, or reorder, because the matching-path code can't be independently audited
The case for Solidus starts here: the technical demonstration is not the same as the complete product. Hyperliquid shows that on-chain perps work. Solidus asks: what does the fully open, fully permissionless, fully trustless version of that look like?
What Solidus is
Solidus is a 16-chain hierarchical L1 targeting Hyperliquid-class performance on fully open infrastructure. Four properties distinguish it:
Open source end-to-end
The consensus, the matching engine (ForumCore), the node software, the SDK — all open source. Anyone can audit, fork, or run the code. This is not negotiable and not implementation-specific; the architecture is designed to be distributable because there is no sequencer to hide.
Permissionless participation
Any validator can join by staking. No allowlist, no operator approval. Four mining hardware classes (ASIC, GPU, CPU, Mobile) can contribute to consensus through POEM (Proof of Entropy Merge) without crowding each other out of economic viability.
Non-custodial asset security
User assets live in their own wallets until they're committed to a trade. Margin is held in on-chain vaults that are mathematically constrained — the protocol, not the operator, enforces settlement. There is no custodial bridge, no centrally-held user balance ledger, no off-chain state that the operator could lose or manipulate.
Sub-second finality without fixed parameters
Solidus uses DAGKNIGHT cascade voting to achieve ~100ms finality. Unlike fixed-parameter protocols, DAGKNIGHT's confirmation requirement adapts to network conditions — fast when the network is calm, more conservative when it's under stress. A dedicated post walks through DAGKNIGHT's adaptive-k mechanism.
The 16-chain architecture
Most L1s are a single chain. Solidus is sixteen parallel chains organized in a three-level hierarchy:
- 1 Prime chain — the coordination layer, handles cross-chain settlement and global state
- 3 Region chains — geographic sharding for latency locality
- 12 Zone chains — four mining algorithms × three regions; transactions bucket into the closest zone
The hierarchy is what allows 16-way parallel execution without the coordination overhead collapsing throughput. Transactions land on the zone closest to them by region and mining class; heavy local traffic doesn't starve other zones; the Prime chain finalizes the global ordering.
ForumCore — the perps engine
ForumCore is Solidus's on-chain order matching engine. Not a fork of Hyperliquid's (which is closed source anyway); a from-scratch design that:
- Matches limit orders with deterministic, verifiable ordering (no operator-level reordering)
- Settles margin and funding on-chain every epoch
- Supports isolated and cross margin modes
- Exposes order book and trade streams through the node's standard RPC
ForumCore is built to be drop-in compatible with existing trading infrastructure — CEX-style API shapes for order submission, WebSocket streams for market data — so traders and market makers don't rewrite their connectivity layer. The difference is that the matching happens on a permissionless L1 rather than an operator's server.
Where Solidus stands today versus Hyperliquid, dYdX, GMX, Aevo
The current perps landscape at a glance:
| Solidus | Hyperliquid | dYdX v4 | GMX | Aevo | |
|---|---|---|---|---|---|
| Source available | Yes (full stack) | Node open, matching engine closed | Yes (full v4-chain open) | Yes (contracts) | Partial |
| Permissionless validators | Yes | KYC/KYB Foundation-gated | Yes (Cosmos) | N/A (contracts on L1) | No (centralized) |
| Target finality | ~100ms (DAGKNIGHT) | ~1s | ~1s | L1 finality | ~1s |
| Matching model | On-chain order book (ForumCore) | On-chain order book | On-chain order book | GM pools + GLV (v2) | Off-chain book, on-chain settle |
| Consensus | DAGKNIGHT + POEM (hybrid) | Custom Tendermint variant | Tendermint (Cosmos) | Ethereum / Arbitrum | Roll-up |
| Asset custody | Non-custodial | Non-custodial | Non-custodial | Non-custodial | Non-custodial |
| Mining hardware classes | ASIC + GPU + CPU + Mobile/TEE | N/A | N/A | N/A | N/A |
Each project solves a different piece. Hyperliquid trades openness for performance. dYdX trades performance for Cosmos-level openness. GMX trades matching quality for AMM simplicity. Aevo trades on-chain purity for off-chain order book performance.
Solidus's bet is that DAGKNIGHT + POEM + 16-chain parallelism closes the performance gap without the centralization trade.
Why multi-algorithm mining matters for the perps thesis
Most L1s secure themselves with a single consensus mechanism: PoS (Ethereum, Solana), PoW on one algorithm (Bitcoin, Monero), or BFT variants (Cosmos chains). Single-mechanism security means the attack surface is uniform — whoever dominates that one mechanism controls the chain.
POEM spreads security across four hardware classes:
- SHA-256 (ASIC) — industrial mining facilities
- ProgPoW (GPU) — gaming PCs and rendering farms
- RandomX-Q (CPU) — server and desktop CPUs
- Panthera (Mobile / TEE) — phones, tablets, and trusted-execution-environment devices
Each class contributes entropy to block ordering through POEM. No one class dominates; all four are priced into block production such that none can be economically attacked without attacking the others simultaneously. A 51% attack on SHA-256 alone does not compromise the chain; the three other classes continue to contribute valid entropy.
For a perps exchange specifically, this is important because consensus attacks are the attack vector for reorgs, and reorgs are the attack vector for trade manipulation. Hardening consensus against single-class attacks hardens the exchange against coordinated trade reversal.
What Solidus needs to prove
A realistic post should name the unshipped parts:
- Production volume. Hyperliquid has billions of dollars in monthly volume. Solidus's track record is early. Volume and liquidity come from sustained market-maker participation, which follows infrastructure maturity.
- Liquidity depth. Order books get thin without market makers. Solidus has to earn liquidity provider trust before volume arrives.
- Ecosystem of dApps. Hyperliquid has a growing application layer. Solidus needs the same to become more than just a perp venue.
- Developer tooling parity. SDKs, RPC stability, block explorer, indexing infrastructure — all have to reach CEX-developer-expectation quality.
These are solvable problems. They are also not solved yet. The thesis is that given the architectural foundation is right, the ecosystem follows — and the architecture is what determines whether decentralization is compatible with CEX-performance at all.
When Solidus is the right bet
Solidus is compelling if:
- You want the performance of Hyperliquid with open-source code you can audit
- You want permissionless validator participation — no whitelist, no operator approval
- You want multi-hardware-class security that doesn't concentrate in a single mining mechanism
- You're building a perps venue, a DEX, or an application that needs sub-second finality on trustless infrastructure
Solidus is not the right bet if:
- You need the battle-tested liquidity of existing venues today
- You're running a workload that doesn't benefit from sub-second finality
- You prefer the simplicity of a Cosmos-style single-chain deployment
Next up in the Solidus silo
- Sub-100ms Finality via DAGKNIGHT Cascade Voting — already live
- POEM Mining Explained: Four Algorithms, No Cross-Class Competition — upcoming
- ForumCore: Designing an On-Chain Order Matching Engine — upcoming
- The 16-Chain Hierarchy: Prime, Regions, Zones — upcoming
Or go straight to the Solidus product page for the testnet and architecture details.