Stop paying for AI.
Plug in your GPU instead.
The Grid turns your idle GPU time into credits for model APIs.
Share when your machine is free.
Use the network when you need inference.
Run models for free, or cash out your earnings.
Anywhere a host joins.
We don't own GPUs and we don't run a data center. Every node on the map is contributed by someone. Tokyo is where the project started — the rest grows wherever a host plugs in.
Models available
Open-source weights, served on the network. Pricing in USD, per-token where applicable.
| Model | Type | Price | Status |
|---|---|---|---|
qwen3.5-72b-instruct | chat | $0.30/1K | live |
llama-4-70b-instruct | chat | $0.28/1K | live |
gemma-3-27b-it | chat | $0.18/1K | live |
deepseek-v4 | chat | $1.10/1K | soon |
qwen3.5-vl-72b | vision | $0.40/1K | live |
pixtral-12b | vision | $0.15/1K | live |
whisper-large-v3 | audio | $0.04/min | live |
kokoro-tts-v1 | tts | $0.08/1K | live |
flux-schnell | image | $0.50/img | live |
bge-reranker-v2-m3 | rerank | $0.02/1K | live |
Got a spare GPU? Join Grid. Use the network for free.
Install OpenAlchemy Engine on any GPU you have — from 8 GB to 80 GB. Grid matches your VRAM to the right models: small GPUs run embeddings and small chat, large GPUs run 70B+. Earn credits as the network uses your GPU. Spend them on your own inference for free, or cash out at 70% revenue share. No contracts.