Network online

Open GPU Infrastructure

The open AI network anyone can plug into. Send OpenAI-compatible requests, route them across a global mesh of everyday GPUs, and earn Season 0 points for every task your hardware serves. No token required. Pay per token of inference. Earn points for spare cycles. OpenAI-compatible, so your existing client just works.

Start providing View quickstart

Latency

< 200ms

p50 first token

Throughput

60 tok/s

consumer GPU

Compat

OpenAI

/v1/chat/completions

Rewards

GPU Points

Season 0 · live

scroll ↓

The compute already exists

The hardware is already out there. It was just sitting idle.

Idle24 GB · 82 TFLOPS

RTX 4090

Idle in a gaming rig between sessions.

Idle40-core GPU · 400 GB/s

M3 Max

Asleep inside a closed laptop.

Idle12 GB · 13 TFLOPS

RTX 3060

Spare cycles between renders.

One request, end to end

Send a prompt. The mesh answers.

01 · Request

02 · Route

03 · Serve

04 · Earn

POST · /v1/chat/completions live

←

Any GPU · any model

Thousands of machines. One network.

Quickstart

One line to start providing. No git, no clone, no setup.

Run a node

gpunetwork.xyz

$ curl -fsSL https://gpunetwork.xyz/install | sh

Installs Ollama + Llama, registers with the orchestrator, starts serving inference. Earns Season 0 points per completed request.

Call the API

gpunetwork.xyz/v1/chat/completions

$ curl https://gpunetwork.xyz/v1/chat/completions \
  -H "Authorization: Bearer dgi_xxx" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpu-mesh","stream":true,"messages":[{"role":"user","content":"hello mesh"}]}'

OpenAI-compatible. Stream SSE chunks back the same shape every SDK already knows.

Live · updates every 4s

The mesh is already moving.

Tasks served

Nodes online

API keys issued