Skip to main content
Gamers contribute incidentally. GPU contributors run hardware specifically to participate in the network — a machine left on for the job, rather than a gaming PC that happens to be idle. Different goals, different optimisation.

​
Who this is for

  • People who already own RTX hardware and want to put it to work instead of leaving it idle.
  • Power users with 24+ GB VRAM cards, prosumer workstations, or ML rigs that sit unused outside business hours.
  • Small operators with 2–10 GPUs looking for stable side income.

​
What makes the economics work

  • Pay scales with demand for AI compute. As API usage and partnership volume grow, so does what there is to pay out. There is no difficulty curve working against you.
  • Nothing to tune. No algorithm selection, no protocol forks to chase, no per-job configuration. The AI App handles dispatch.
  • GUSD pays per workload completed — predictable, internal, no exposure to coin-price volatility while you decide what to do with it.
  • GHXP rewards consistency. Points accrue per minute of uptime, not per unit of throughput, so a machine that stays online steadily climbs higher in each monthly season ranking.

​
Optimisation playbook

Maximize uptime. Most pay comes from being online when a workload arrives. Configure auto-start, disable sleep, monitor stability. Pick the right cards. 24 GB+ VRAM unlocks the highest-paying tiers (large LLMs, video gen). RTX 4090 / 5090 / Blackwell pro cards are the sweet spot for new investment. Hold GHX for the boost. Holding GHX raises your account level in the app, and the level is what accelerates GHXP accrual. It also unlocks the prompt booster on Text-to-Image. It’s free, it stacks, and you don’t lock the GHX — hold it in MetaMask or in your in-app GHX wallet and run the app. More points means a higher place in the monthly season ranking. Run multiple machines on one account. GHXP and earnings consolidate per account, not per machine — so several PCs build one points balance instead of several small ones. Lock points every month. A season runs monthly and you can commit up to 80% of the GHXP you’ve accumulated. Points you never lock don’t place you anywhere, and every place in the ranking carries a defined GHX reward.

​
The numbers conversation

There is no fixed or promised rate on the earning side. Per-machine earnings depend on demand, GPU class, and how much of the time the machine is online and free. The honest framing:
  • A 16 GB Ada card with 64 GB of RAM sits at the entry earning tier and will earn modestly when network demand is moderate.
  • A 24 GB+ card draws from the full workload pool, including the top-paying tiers, and with high uptime will earn meaningfully more.
  • All of it scales with platform API and partnership volume — which is the variable you’re betting on.
The platform publishes monthly community updates on Medium with workload volume signals. That’s the right place to read the room before scaling up.

​
What to watch out for

  • Electricity cost. At peak tariffs this is the difference between a margin and none. Schedule earning during off-peak hours if your tariff is time-of-use.
  • Driver stability. Long-running inference exposes driver bugs. Stick to NVIDIA Studio drivers; report crashes to the support channel.
  • Heat / hardware longevity. AI inference is GPU-intensive. Good airflow matters. Undervolting is supported and recommended for 24/7 operation.

Earning model

What a completed job pays, and the three ways the balance leaves your account.

Activity points and seasons

How points accrue, and how the monthly ranking is settled.