Network
Workers and earning
Contribute a GPU to the swarm and earn per job. Native workers wrap a local model for maximum throughput; browser workers run in a tab with zero install.
Worker classes
Run a native worker
Install Ollama, pull a model, then start the worker. It opens a WebSocket to the orchestrator and begins accepting jobs.
# 1. install a model runtime and pull a model
ollama pull qwen2.5:27b
# 2. start the NoviQ worker (connects to the orchestrator)
npx @noviq/worker \
--mode native \
--model qwen2.5:27b \
--token $NOVIQ_WORKER_TOKENWorker tokens
Run a browser worker
Open the earn page on a WebGPU-capable browser. The model loads into the tab and the machine becomes a worker for as long as the tab is open - nothing to install, nothing left behind when you close it.
# just open it in a WebGPU browser
https://noviqai.xyz/earnHow you get paid
Earnings are per job, based on the worker class and the completion size. Native jobs pay a base rate plus an amount per thousand completion tokens; browser jobs pay a smaller base plus a smaller per-token amount. The orchestrator records the amount for each job from token counts only.
Receipt-gated payouts
Paid for real, correct work
