Overview

CompressNode is a decentralized AI inference network that lets anyone with a consumer GPU serve AI models and earn ETH. Using GGUF 4-bit quantization (branded as TurboQuant), large language models that normally require expensive data center hardware can run on consumer GPUs like the RTX 4060 - with near-zero accuracy loss.

The CompressNode network is built on three pillars:

• TurboQuant Compression - GGUF 4-bit quantization reduces model memory by 6x while preserving output quality
• Permissionless Node Registration - anyone can join and start serving inference, no staking required
• Math-Based Verification - on-chain challenge-response system proves nodes are running authentic models
• ETH Payment Economy - all payments flow through native ETH on Base, with 95% going to operators
• On-Chain Reputation - nodes build reputation through successful job completion and verification

Two types of participants interact with the network:

• Node Operators - GPU owners who run the CompressNode daemon, serve inference requests, and earn ETH
• Users - developers and applications that submit inference jobs and pay ETH for AI responses

CompressNode is deployed on Base, Coinbase's Layer 2 network. Base provides sub-cent transaction fees, fast finality, and Ethereum's security guarantees - making it ideal for high-frequency inference payments that would be prohibitively expensive on Ethereum mainnet.