This page sets the two branches side by side by architectural property, not by score. Neither column is the winner: the two branches make different trade-offs, and which one fits depends on what you intend to build. Every cell comes from that branch's own source — no cell is filled in on the other's behalf.

Property	9Chain A1	9Chain C1
Consensus family	Snowman — repeated random sampling of the signing set	CometBFT — every signer votes on every block
Block cadence	On demand — a block exists only when someone transacts	Fixed cadence — blocks are produced even when nothing happens
Chains in the network	Many L1s, created on request	A single chain, everything shares it
Transaction index	Not built in — a separate indexer has to be run	Built into the node
Transaction fees	An elastic fee market, sitting at its floor while the network is quiet	No fee charged — the minimum gas price is set to zero
On-chain voting	None — parameters change by operator action	A governance module: proposals, deposits, on-chain voting
Compliance in the protocol	None at the protocol layer — it sits outside the chain	Enforced in the protocol, with a gate contracts can call into
Cross-chain messaging	An internal messaging mechanism — not switched on	The standard module is compiled in, with no working route

Five places worth reading more closely than the rest, because these are where a tidy table is easiest to read as something it does not say.

One — the two block heights cannot be subtracted from one another. A fixed cadence keeps block time predictable but grows the chain even while it sits idle. On-demand production costs nothing when idle — but it takes away silence as a health signal: a branch that has stalled and a branch that is merely quiet look identical. That is why no page of the project is allowed to raise an alarm when the on-demand branch has produced no new blocks.

Two — "free" and "paid" are closer together than they look. The branch that charges no fee has to hold back abuse some other way (a per-account gas quota), and the zero floor is a setting on machines the project runs, not a guarantee of the protocol. The other branch keeps an elastic fee market, which sits exactly at its floor while the network is quiet. So the real distance between them is far narrower than the word free suggests.

Three — what protocol-level compliance buys, and what it costs. Rules enforced below the contract layer apply to every transfer on the chain, and no contract can route around them. But they apply only to transfers on the chain, and whoever can upgrade the chain can still change the rules. The cost, stated plainly: that branch is permissioned — someone decides who may transact. The other branch pushes that decision up to whatever runs on top.

Four — an on-chain vote, at this stage, records a decision more than it makes one. A governance module leaves a public record of every parameter change, at the cost of taking days and needing a quorum. But on a test network where most of the voting power sits with the project itself, that ballot is mostly a set of minutes. Operator action takes effect immediately and leaves no such record — faster, and dimmer.

Five — interoperability is an unpassed gate on both sides. One branch has the standard module compiled in, but a working route also needs a counterparty chain and someone running a relayer; on the other, the equivalent mechanism is not switched on. Neither is moving assets in or out. Treat both as unproven here.

Diagram

flowchart TB Q{"What do you need first?"} Q -->|"many separate chains, one per workload"| A["The branch that creates many L1s on request"] Q -->|"compliance rules contracts cannot skip"| C["The branch that enforces in the protocol"] Q -->|"a public record for every parameter change"| C Q -->|"absolute finality the moment the round closes"| C Q -->|"a signing set that scales up"| A A --> G["Both are test networks: neither has been load-tested, neither is a commitment to run for real"] C --> G

Figure 12 — Read by need, the table above produces this shape. But every path leads to the same final box, and that box is the most important thing on this page.

One thing this table deliberately lacks: any operating figure. No time to finality, no throughput, no count of machines signing. The reason is on Measurement and the ballot — and it is not laziness but the fact that a frozen number would have both branches judged by a measurement nobody can reproduce.

In short: the two columns above are trade-offs, not scores — and the line worth remembering is not in the table at all: neither branch has been load-tested, and neither is a commitment to run for real. Anyone who reads this table and immediately knows which side wins is almost certainly skipping the trade-off column of the side they just picked.

→ Next: Measurement and the ballot — The three gates standing before the ballot