Resource network

GPU

GPU capacity for training, rendering and serving — drawn from the machines in your gram, or from hardware stood up inside your own walls.

Metered in VRAM-hours

Where it can run

The same resource, the same meter, three answers to “whose machine is this?” You choose per workload, and you can move a workload inward when the rules tighten.

On your device

Available

Inside your gram

Available

On a stranger’s machine — not available

Available

What is proven — and what is not

Each node benchmarks itself and signs the result with its own key. Where a benchmark does not exist yet, the node says so and emits nothing in its place. So do we.

Signed metric

gpu.throughput

Unit

name · VRAM · utilisation

Partially measured

How it is measured

The node reports its GPU inventory — model, total VRAM, current utilisation — read from the vendor tooling on the machine, and signs it.

What this does not tell you

This is an inventory, not a benchmark. A signed GPU throughput number is configured and not yet benchmarked, and the node emits no fake one in its place. Detection currently depends on NVIDIA tooling. When we can measure it, it will appear here signed, and not before.

Why GPU is its own network

The industry sells GPUs by model name and lets you assume the performance. We would rather hand you an inventory we can sign and tell you plainly which benchmark is still missing.

When it breaks

Free on your own hardware. The meter only runs when you consume someone else’s.

There is no 24/7 operations centre, because we do not employ one. Instead a node that cannot prove it is healthy is evicted from the ring rather than quietly serving your work. Fail-closed, not fail-silent.

How you leave

Bring hardware onto your own site, funded upfront or financed. It is yours; we operate it.

How it is delivered

Every resource above reaches your workload through the same substrate, whichever ring it was drawn from.

Kubernetes, on every node

Each node runs k3s. One orchestration layer schedules all seven resources, so a workload moves between rings without being rewritten.

Virtual servers, when containers will not do

Workloads that cannot be containerised run as virtual machines on the same cluster, through the open-source KubeVirt project. Same scheduler, same meter.

Nothing proprietary in the exit

Standard containers, standard VMs, content-addressed objects, exportable receipts. The cost of leaving is the reason to trust the platform.

What you are billed for

GPU is metered in VRAM-hours. Each unit of work produces a receipt naming the node that performed it and the price it was charged at, on the one ledger the whole platform shares. We publish no hourly rate table on this page, because a price is meaningless without the signed unit it is counting.

See how we price compute →

Tell us what the workload is.

We will tell you which ring it belongs in, what it will be metered in, and what we have not measured yet. We scope before you pay.

What a machine actually did

Not a specification, not a vendor sheet. The best verified result any node on this network has signed for gpu — and, below it, everything else that node put inside the same signature.

gpu.throughput

67.23 TFLOP/s

Measured
Sep 13, 2026, 05:49 PM
Signed by
did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54
Payload sha256
6/UixcPV/vQ4i9V34sYqb7km5qWZ/X3HFJ5l2aDILXA=

What the headline hides

A benchmark reports one number and signs several. These travelled inside the same signature, so they are as verifiable as the figure above — and they are usually the ones that decide whether the machine suits your work.

dtype

fp16

dtype

Signed by the node. We have not written an explanation for this field yet, and would rather show it unexplained than invent one.

elapsed_ms

102

elapsed_ms

Signed by the node. We have not written an explanation for this field yet, and would rather show it unexplained than invent one.

iterations

50

iterations

Signed by the node. We have not written an explanation for this field yet, and would rather show it unexplained than invent one.

matrix

4,096

matrix

Signed by the node. We have not written an explanation for this field yet, and would rather show it unexplained than invent one.

mem_total_mb

12,288

mem_total_mb

Signed by the node. We have not written an explanation for this field yet, and would rather show it unexplained than invent one.

Method

torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot

method

What was actually run, in the words of the node that ran it.

name

NVIDIA GeForce RTX 3080 Ti

name

Signed by the node. We have not written an explanation for this field yet, and would rather show it unexplained than invent one.

source

epn-gpu-bench job (docker.io/pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime)

source

Signed by the node. We have not written an explanation for this field yet, and would rather show it unexplained than invent one.

What we still cannot tell you