For GPU providers

Your capacity.
Your choice to contribute.

IdleGPU is exploring a controlled way for invited GPU owners to process approved work and earn an agreed share for accepted results.

Hardware-based earning tiers

Different GPUs.
Different earning potential.

We’re designing participation around a range of GPU capabilities. Compatible entry-level hardware can handle smaller jobs; more capable hardware can qualify for higher-rate work when its measured performance justifies it.

STARTER

Everyday capacity

Entry-level and everyday devices, matched to approved jobs they can complete correctly.

PERFORMANCE

More useful throughput

Midrange hardware with tested capacity to deliver suitable batches faster or handle more demanding jobs.

ADVANCED

Greater workload capacity

High-performance and higher-memory systems where compatible workloads benefit from the extra capability.

Having a GPU does not guarantee admission. Software compatibility, usable memory, correctness and reliable delivery all matter. Earnings require paying demand and accepted work.

Explore tier rates →
CONTROL

Availability you choose

Explicit opt-in and manual pause. Planned memory and execution limits should make participation understandable and controllable.

WORK

Approved tasks only

Fixed models and bounded inputs. No customer-supplied code, unrestricted remote rental or public inbound worker service.

VALUE

Accepted work matters

Compensation would follow validated output—not an unverified claim of GPU time. Failed and disputed work needs a defined review process.

Participation, step by step.

Register interest

Describe the device and availability without sending credentials or confidential details.

Validate the device

Record hardware, software, correctness and sustained performance with an invited test.

Choose availability

Opt in to approved work. Outbound polling and expiring leases handle disconnects.

Review accepted work

Inspect completed records and the agreed compensation before supervised payouts begin.

Provider economics

Useful earnings need
useful demand.

A connected GPU does not automatically receive paying jobs. Net benefit depends on accepted work, electricity, availability, cooling, bandwidth and wear.

The pilot must measure these costs and failed-work overhead before quoting income. No minimum earnings or utilisation is promised.

Explore provider tiers →