Open-source prototype · View on GitHub

The compute network where AI work secures the chain

Noetis replaces wasteful hash mining with distributed LLM inference. Anyone with a GPU contributes real compute — and earns rewards for verified AI output.

See how it works Explore the code
Hover a node

Click any node in the network to learn what it does.

COORDINATOR consensus + rewards GPU-01 GPU-02 GPU-03 GPU-04 GPU-05 GPU-06 YOU Hover nodes · click to pin · live TCP mesh on port 9600
$50B+Annual AI compute spend
90%GPU time sits idle
0%Value from Bitcoin mining
PoIProof-of-Inference

The problem

Two broken models holding AI back

Centralized clouds concentrate power and cost. Blockchains burn energy on useless hashes. Neither model scales open AI for everyone.

☁

Centralized AI is a bottleneck

A handful of providers control inference. Costs rise, access narrows, and millions of idle GPUs contribute nothing.

⛏

Crypto mining wastes compute

Proof-of-Work burns electricity on SHA-256 hashes with zero external value. That hardware could run real AI.

Our solution

Useful work replaces wasteful mining

Noetis coordinates a mesh of independent nodes that run real LLM inference locally. Consensus verifies outputs. Contributors earn MLC tokens for correct, fast responses.

⚡

Proof-of-Inference

Each block records a verified AI task — not a meaningless hash puzzle. Security from redundant inference and majority consensus.

🌐

Open compute mesh

Laptops, desktops, and servers join via simple worker software. No single model host. Compute stays distributed.

How it works

Interactive protocol flow

Click a step or hover nodes to explore how a prompt becomes verified AI work and MLC rewards.

User submits a prompt

A client sends a query to the coordinator over a lightweight TCP socket protocol.

📂 connection-layer source ⚙ Worker setup guide 📄 Protocol docs

Why Noetis

Built different from day one

Bitcoin / PoWCentralized APINoetis
Work performedSHA-256 hashesVendor-hosted inferenceDistributed LLM inference
External valueNoneHigh (but gated)High + open access
Who can contributeASIC minersNobodyAny GPU / CPU owner
Incentive modelBlock rewardsSubscription feesMLC for verified work
StatusProductionProductionWorking prototype

Roadmap

Where we're headed

The connection layer is live. On-chain settlement and production hardening come next.

Phase 1 · Done

Connection layer

Socket protocol, worker mesh, Ollama inference, consensus, MLC ledger demo.

Phase 2 · Now

Network growth

Onboard early contributors, stress-test incentives, refine verification.

Phase 3

On-chain settlement

MLC token economics on a blockchain layer. Staking, slashing, Sybil resistance.

Phase 4

Global mesh

Open inference marketplace — cheap, censorship-resistant AI worldwide.

Let's build the open AI compute layer

Early contributors, technical partners, and investors welcome.

Connect on LinkedIn GitHub repo