Cloud Run
Serverless HTTP workloads — per-request billing (always-on reserved equivalent)
$41.5k/yearly
≈ 113.8M tokens/day
- vCPU
- $33.9k/yr
- RAM
- $7.6k/yr
Limited pilot · the dollar amounts below are indicative projections of what a device could earn if paid services launch — not actual payouts. No cash is paid during the pilot. Full disclaimer.
Live · computed from devices enrolled right now
What the network could be earning.
Everything enrolled right now — 7 devices · 64 vCPUs · 138 GB RAM · 6 GPUs — adds up to an estimated $80.6k/year of cloud-equivalent compute — minting 220.8M compute tokens/day at full potential.
GKE Node
Reserved Kubernetes node slot (e2-equivalent, us-central1 list pricing)
$24.3k/yearly
≈ 66.6M tokens/day
- vCPU
- $18.7k/yr
- RAM
- $5.7k/yr
AI Inference (GPU)
On-device inference accelerated by the device's own GPU — LLMs, image generation, transcription
$80.6k/yearly
≈ 220.8M tokens/day
- vCPU
- $55.7k/yr
- RAM
- $10.2k/yr
- GPU
- $14.7k/yr
The pool keeps growing
Earning potential of everything enrolled, sampled hourly and priced at today’s rates — the curve is capacity growth, not price drift.
⚡ Test your own device
Pick something like yours, then drag the sliders. Values update in place.
Indicative only, not actual payouts. Omnira is in a limited pilot and pays no cash today; these figures are illustrative projections of what a device might earn if and when paid services become generally available — not money owed, a promise of payment, or a guarantee of future earnings. Estimates are benchmarked vs 2026 public cloud list prices at 70% utilisation; actual earnings, if any, depend on paid demand, final pricing, your device’s real contribution, fees, taxes, and payout verification. Rate sources (list-price snapshot 2026-07-20): Cloud Run · GCE e2 · Vertex AI · GCE GPUs (T4/A100) · AWS G5 (A10G).