The AI Revolution Will Not Be Centralized
JimboMesh is a peer-to-peer AI compute network. Run models on your hardware. Sell your GPU power. Buy inference cheap. No cloud giants. No permission needed.
The Platform is Live
Sign up free — GitHub, Google, Microsoft, or Apple login.
Dashboard
Manage your account, monitor your Hollers, track Moonshine earnings, and view network stats — all in one place.
Playground
Chat with AI models running on real GPUs — no setup required. WebRTC peer-to-peer. Your data never touches our servers.
API Keys
Generate API keys, connect your IDE, point your apps at the mesh. One key, hundreds of models.
Holler Fleet
Monitor your Holler nodes, track uptime, manage connections, and watch your Moonshine flow in real time.
See It in Action
Take the Back Road to AI Freedom
The cloud giants control the highway. JimboMesh is the back road — decentralized, community-owned, and open to anyone with a GPU.
Ready to join the mesh?
Fire Up a HollerQuick Start
Five minutes from download to a running local AI server.
git clone https://github.com/IngressTechnology/jimbomesh-holler-server.git
cd jimbomesh-holler-server
.\setup.ps1
# Open http://localhost:1920/admin
git clone https://github.com/IngressTechnology/jimbomesh-holler-server.git
cd jimbomesh-holler-server
./setup.sh
# Open http://localhost:1920/admin
git clone https://github.com/IngressTechnology/jimbomesh-holler-server.git
cd jimbomesh-holler-server
./setup.sh
# Open http://localhost:1920/admin
Got a GPU Collecting Dust?
Your $3,000 graphics card sits idle 22 hours a day. Fire up a Holler — run local AI inference, document RAG, model marketplace, admin dashboard. All on YOUR hardware. Connect to the Mesh and earn Moonshine. Your RTX 4090 has been freeloading long enough.
Fire Up a Holler →Need Cheap AI Inference?
You're paying OpenAI $20 per million tokens. Or waitlisted for GPU instances that cost more than your apartment. JimboMesh gives you one API endpoint backed by a distributed network of real GPUs. Same OpenAI-compatible format. Change one base URL. Quarter the price.
Start Building →The Marketplace
Providers earn Moonshine. Consumers spend Moonshine. The network grows. No middleman taking 70%. No cloud provider deciding your GPU isn't good enough. It's buying moonshine instead of top-shelf bourbon. Same buzz. Quarter the price.
Enter the Marketplace →How It Works
Your data never touches our servers. That's the whole point.
- 1 You request AI inference through the JimboMesh API or Playground
- 2 The platform handles signaling — matches you with a Holler that has your model
- 3 WebRTC connects you directly to the GPU. Peer-to-peer. No middleman.
- 4 Your inference data goes straight from your browser to the Holler and back. We never see it.
WebRTC Peer-to-Peer
The platform handles signaling and billing.
Your inference data goes peer-to-peer.
This is the #1 technical differentiator vs. every other AI platform. No proxy. No relay. No data harvesting. Just you and a GPU.
Try It Now
The Playground
Chat with AI models running on real GPUs — no setup required. Three tabs, real Hollers, WebRTC peer-to-peer connections. Pick a model, start talking.
- Browser-based — no install, no API key needed to try
- Connected to real Holler nodes on the mesh
- WebRTC direct connection — your data never touches our servers
- Multiple models available
Kinfolk
Private Holler groups — like a LAN party for AI.
Share GPU compute with your friends for free. Set up a private Holler group — your machines, your models, your rules. No Moonshine required. It's just your people and their hardware.
Outgrow your group? Join the marketplace and start earning from the whole mesh. But Kinfolk comes first — because the best networks start with the people you trust.
Why JimboMesh?
- → Because 200 million gaming PCs sit idle 22 hours a day
- → Because three companies control AI compute pricing
- → Because the best technology gets built in garages, not boardrooms
- → Because moonshine always finds a way
🥃 Alpha is LIVE
Join the First Operators and Builders Shaping the Network
The platform is running. The mesh is growing. Early operators are earning Moonshine and early developers are building on the API. Get in now and help shape what comes next.
Join the Alpha →"Nothing works until it works once. After that, it can always work again."