GPU Provider Income Tracker

Realised provider income is decided by two variables the marketplaces downplay: utilisation, and whether you are paid in cash or in a token you have to manage. This tool models both across cash marketplaces and token networks, with a live Morpheus column nobody else publishes. It is a parametric model, not an earnings oracle. The full reasoning is in Where Should You Deploy Your GPU?

GPU Provider Income Tracker

This is a parametric model, not a live earnings oracle. Utilisation cannot be observed per user, so you set it as an assumption. The tool never claims to know what you will earn.

Marketplace rates last reviewed 3 Aug 2026 · token prices fetched live · figures point-in-time as of mid-2026

GPU model Electricity $0.20/kWh Utilisation (your assumption) 30% Home consumer cards typically run 20-50%; datacentre cards 70-95%. Token price multiplier 1.0× Scales token-network income only. 1.0× = mid-2026 token prices.

Subtract hardware amortisation Token prices: loading…

Cash payouts

fiat / USDC · no token to manage

Vast.ai fiat —

Rolling, no minimum, fiat or crypto · Plug-and-play

RunPod Community fiat —

Rolling, fiat · Plug-and-play

Akash (settle USDC) USDC —

Per-lease, on-chain · Kubernetes provider stack

Token payouts

price exposure · model assumes sell-on-receipt

io.net IO —

IO price risk until the Incentive Dynamic Engine dollar-target payouts go live (planned Q2 2026). Consumer cards mostly earn thin idle block rewards; job rate shown is when actually booked.

Render RENDER ⏱ delay —

RENDER price risk plus a 7-10 day payment delay, so you cannot sell on receipt. Best fit for high-VRAM cards.

Aethir ATH self-reported —

Self-reported, not independently audited. Enterprise 8-GPU H100-class nodes only; US residents cannot monetise.

Token rows assume you reprice and sell on receipt, with ~2.5% slippage applied. Hold the token instead and it becomes a separate investment position, not compute income.

Morpheus — live on-chain network reality

OYM indexer · 2026-08-25

~3116

MOR/day settling to all providers (30d)

~3011

MOR/day allocated to the compute pillar

addresses have ever settled a session

17/50

of the latest sessions from one address

This is what the whole network settles, not your expected share. Supply is close to a monopoly. The pillar is not fully distributed and a new provider fights one incumbent for a slice of a thin pot. At ~$2.00 per MOR the entire ~3116 MOR/day is roughly ~$6232/day across all providers combined. Deploy here for conviction, not income.

Net monthly = (host-kept rate − electricity − amortisation) × utilised hours, floored at zero, rounded hard. Outputs show the low-to-high host-rate band. Bar length is the share of a full bar, where a full bar is the best platform at 100% utilisation and today's token prices, so the bars grow with utilisation and a token row caps once it beats that best case. Aethir and Venice figures are self-reported or undisclosed; Venice is excluded because you cannot provide to it. The article Where Should You Deploy Your GPU? is this tool's backend.