A proposed pricing framework for validation—not an active offer or an earnings guarantee. All amounts exclude any applicable GST; tax treatment is pending review.
Better hardware can qualify for higher productive-hour rates when benchmarked throughput, memory and reliability justify it. Idle time, failed work and unsupported workloads do not create earnings.
Everyday and entry-level GPUs, where compatible with the approved workload. Placement depends on usable memory, software support and measured processing capacity.
Customer planning rate
A$0.45 / hour provider gross compensation
A$0.45 retained before validation, hosting, payment and support costs.
Modern midrange GPUs that deliver a measured step up in throughput or workload capacity. The specific model, memory and configuration are assessed together.
Customer planning rate
A$0.80 / hour provider gross compensation
A$0.70 retained before direct operating costs.
High-performance or higher-memory GPUs and compatible AI systems. Higher rates apply when approved jobs can use the additional capability—not just because the hardware is expensive.
Customer planning rate
A$1.20 / hour provider gross compensation
A$1.20 retained before direct operating costs.
Tier placement is provisional until a compatibility and workload benchmark is completed. A device may qualify differently for different workloads. “Hour” means accepted productive execution attributable to approved work, not connected time. These candidate rates are untested and may change or be withdrawn. Provider returns depend on actual customer demand and acceptance.
Use benchmarked execution, item/token limits and an approved model to quote a fixed batch ceiling. Choose the best all-in price for the workload, not automatically the strongest GPU.
Proposed minimum invoice: A$10 of defined accepted work, aggregated across batches. Requote scope changes. No unapproved spend above the ceiling.
Bill accepted output only. Provider telemetry is insufficient evidence; calibrated execution and trusted validation are required. Slowdowns cannot inflate a fixed quote.
Proposed provider payout: monthly, above A$20, after reconciliation. Carry smaller balances forward. Final timing, refund and dispute rules require an agreement.
No prepaid credits, deposit, invoice or real payout is currently activated. Model loading, retries, validation and fees must be included in the quoted economics, with failure costs assigned explicitly—not silently charged to customers.
Illustrative calculation only. Change paid accepted hours, power and electricity price. Connected or available hours are not paid hours. Actual power, idle consumption, cooling, bandwidth and wear need measurement.
The calculator assumes paid work is available at the selected tier. It does not forecast demand or include every expense. Net after electricity is not profit or take-home income.
A faster tier may finish a job sooner and cost less overall, or a small task may run better economically on a cheaper GPU or CPU. Tier admission depends on workload-specific benchmarks and reliable delivery.
Customers already have rental and managed API alternatives. We need to prove convenience, output quality and all-in value; no guaranteed discount is claimed.
Join early access →