FALLEN-8 CLOUD · PRICING

One product. Three meters. No tiers.

Fallen-8 Cloud is not sold in tiers, because there is only one thing to sell: the whole engine. Every customer gets the full query surface, C# fragments included. What differs between a small customer and a large one is consumption, and consumption is something you measure rather than something you package.

Fallen-8 Cloud is in development. Nothing on this page is billable yet, there is no checkout, and no rates are published: they are set once the infrastructure they pay for is chosen.

Memory-time
PER GB OF RAM, PER HOUR

Your graph lives in memory, so the memory it holds for the hours it runs is the main thing you pay for.

  • Measured on the instance, never on your query volume
  • Stop the instance and the meter stops
  • Namespaces inside the instance are included, up to a cap we will state
  • Checkpoint storage is included rather than metered
Cores
PER CORE, PER HOUR

The compute Fallen-8 occupies. This is the one line that exists wherever we run it for you, including inside your own account.

  • Buys read concurrency, analytics throughput and ingestion headroom
  • Does not make writes faster, and we will not pretend otherwise: the engine commits writes on one thread per namespace
  • The same unit whether the instance runs on our infrastructure or yours
Sidecar use
PER UNIT, PER SIDECAR

The AI-native parts that cost real compute, charged by what they actually process. Use none of them and pay nothing for them.

  • Embeddings, chat and NL assist: per million tokens
  • Document conversion: per page converted
  • Entity and key-term extraction: per thousand calls
  • Observability is never billed, on principle
  • The agent surface is not billed either, see below
TWO WAYS TO RUN IT

Same product, same engine, different compute.

Where the instance sits is a placement decision, not a price ladder. Sovereign deployment is therefore not an expensive top tier: it is one of two ways any customer can have this.

On our infrastructure
ALL THREE METERS

We supply the compute and operate the instance. You pay for what you occupy.

  • Memory-time, cores and sidecar use
  • A dedicated container per customer, which is what makes full C# safe to offer
  • Managed checkpoints, and restore when you need it
  • F8 Studio included
In your own account
CORES ONLY

We operate Fallen-8 Cloud inside your cloud, your data centre, or an air-gapped network. Your infrastructure bill stays yours, and you pay us for cores.

  • No memory-time and no sidecar charges, because that compute is already yours
  • Sovereign European cloud, on-premises, or fully disconnected
  • Your sidecars, so document text never leaves your account
  • Data processing agreement and a support arrangement alongside it
WHAT IS NOT METERED

The things a graph database should not nickel-and-dime.

Queries, traversals, path finding and analytics
No per-query charge, no per-traversal charge, no per-analytics-run charge. They are bounded by the memory and cores you already pay for, and metering the fastest thing the engine does would be a strange way to price it.
The agent surface
MCP tool calls are not billed. It is a protocol surface over a graph you already pay for, so charging for it would mean metering your queries through the back door, and pricing an agent differently from the identical REST call it replaces.
Observability
Logs, metrics, traces and the dashboards over them are never billed. Charging you to find out what your own instance is doing would be a tax on operating it well.
Namespaces and checkpoints
One instance holds many isolated graphs, each with its own indexes, subgraphs and change feed. They are included up to a cap rather than metered, and the cap is a real number we will publish, because each namespace is a real engine with its own writer thread and "unlimited" would be a lie.
Prices, until they are published
Rates for all three meters will be quoted in EUR, with VAT where it applies, and published before the console opens. Nothing will be charged before you have seen them.
RUNNING IT YOURSELF IS FREE

If you run it, you pay nothing.

Fallen-8 is MIT licensed. Download it, run it, put it into production, keep it there: no fee, no meter, no account, no relationship with us required. That is not a trial and not a tier, it is the licence, and it covers the whole engine including the C# query surface. Nothing above applies to you.

The meters start when we operate an instance for you. That includes the case where the instance runs inside your own cloud account, because we are still the ones running it. The line is who operates it, not where it sits.

Want to try the engine without installing anything? The playground runs Fallen-8 compiled to WebAssembly, entirely inside your browser tab. No account, no server allocated to you, and your data cannot leave your machine because there is nothing to send it to.
Open the playgroundSelf-host for free
FALLEN-8 CLOUD · COMING SOON

Tell us to tell you.

The console is where you will create and manage instances, and it is not live yet. Sign-in, instance management and billing ship together with the hosted offering. Fallen-8 Cloud is how the project sustains itself; the engine stays MIT licensed and free to run yourself either way.

Join the waitlistAbout the console

The waitlist is one email to hosted@fallen-8.com. No form, no stored data on this site.