Skip to content

Operators

Earn from your hardware

GPU or CPU-only — install the node agent and earn from every verified job. On Windows and Linux, the installer provisions the managed execution runtime automatically. Serve inference, host agents, run knowledge base workloads, and fine-tune models.

What you need

GPU: Any NVIDIA or AMD card with recent drivers for full-speed inference.

CPU-only: No GPU required. BitNet 1-bit models run LLM inference on any CPU — even a laptop. Lower barrier to entry, same verified earnings.

Runtime: Windows and Linux installers provision the managed execution runtime automatically. macOS installs the signed node service and local API for control and native streaming.

Network: Stable internet connection. The node agent polls the hub every 10 seconds for new jobs.

OS: Windows 10+ and systemd-based Linux distros are first-class managed-runtime hosts. macOS 13+ is supported for operator control and native streaming.

Setup

One command to install

The same install flow registers the signed node service everywhere. Windows and Linux also bootstrap the managed OCI runtime automatically, so operators are not asked to provision a separate container stack by hand before they can be useful.

Windows

PowerShell (as Administrator)

$ iex ((New-Object System.Net.WebClient).DownloadString('https://api.ryvion.ai/install.ps1'))

macOS

Terminal

$ curl -sSL 'https://api.ryvion.ai/install.sh?platform=macos' | bash

Linux

Terminal

$ curl -sSL https://api.ryvion.ai/install.sh | bash

New

No GPU? No problem.

BitNet enables 1-bit LLM inference that runs entirely on CPU. Any modern computer — including laptops — can serve verified AI workloads and earn from the network.

No GPU required: 1-bit quantized models use integer arithmetic that modern CPUs handle natively.

Lower barrier to entry: Turn any idle machine into an earning node — desktops, laptops, cloud VMs.

Same verified earnings: CPU inference jobs produce signed receipts and earn the same 85% revenue share.

Workloads

What your node can serve

As the platform adds workload types, your node automatically receives more jobs.

Chat inference (streaming + non-streaming)
Text embeddings for RAG and search
Image generation (SDXL Turbo)
Audio transcription (Whisper)
CPU inference via BitNet
Knowledge base embedding jobs
Agent hosting containers
Fine-tuning (QLoRA → GGUF)

Operator economics

Operators keep 85% of revenue from every completed job. Payouts are tied to signed receipts — you earn from verified work, not uptime.

85%

Revenue share per job

T+24h

Earning clearing period

Stripe

Direct bank payouts

Example: 500 chat jobs/day (~1M tokens) earns roughly ~$1.50/month at current inference rates. Agent hosting at $0.50/hr can earn $300+/month per active deployment. CPU-only nodes have zero hardware cost — earn from what you already own.

Get started

Start earning today

Install the node agent in under two minutes. Payouts via Stripe Connect.