NETWORK · the live GPU mesh
THE NETWORK
Every card that lists itself shows up here - who's online, who's idle, and how fast they train each base model. Providers earn on their GPUs; holders stake to take a slice of every job. Live from the coordinator, with a demo snapshot when it's offline.
THE NETWORK · live GPU mesh
GPUs you don't own, waiting for work
Anyone can list a card. The coordinator tracks who's online, idle, and how fast they train each base model. ● demo snapshot - coordinator offline, showing sample GPUs
| worker | gpu | region | runtime | status | models | jobs | seen |
|---|
PROVIDERS · earn on your GPU
Rent out the card while you sleep
List any CUDA, ROCm, or Metal GPU. The coordinator routes training jobs to you by throughput and pays you per run - no identity of the submitter ever reaches your machine. You keep 70% of the job price, 85% if you stake. ● demo earnings
become a provider
$ npm i -g avenia-worker
$ avenia worker benchmark # measures your GPU
$ avenia worker start \
--runtime cuda \
--models aven-qwen-0.5b,aven-llama-1b \
--account acct_you
● registered · idle · waiting for jobs
● you keep 70% (85% staked) of every runtop providers
| provider | jobs | gpu-time | share | earned |
|---|
STAKING · holder vault
Hold the token, earn the network
Stake $AVEN to take a pro-rata slice of the holder pool that every job feeds. Providers who stake also unlock the boosted 85% worker share. ● demo vault - sample position