@DocumentingPRL

Pearl is a decentralized AI inference network that uses proof-of-useful-work mining, where GPUs earn $PRL tokens by serving AI model computation (matrix multiplication) rather than arbitrary hashing.

The network launched mainnet on April 27, 2026, has been live for ~3 months with 261.7M tokens mined, recently joined NVIDIA's Inception program, and is building INT7/floating-point inference infrastructure for open-weight models on enterprise GPUs.

The project addresses a real technical problem (verifiable GPU compute) and has working infrastructure, though it's very early-stage with minimal social presence.

Green flags: Working mainnet launched April 2026 with verifiable on-chain mining activity and published network stats · Joined NVIDIA Inception accelerator program, indicating institutional validation for AI infrastructure approach · Published security post-mortem on first network vulnerability (Jul 11-13), showing transparent engineering practices · Novel technical approach to proof-of-useful-work using AI matrix multiplication with cryptographic verification · Very early stage: 283 followers, 3-month-old account, pre-discovery phase

Red flags: Extremely low social engagement (avg 19 interactions) despite 3 months of mainnet operation · No website data available, making independent verification difficult · Only one Tier B scout tracking it with moderate track record (50/100) · Complex technical claims about GPU verification and INT7 quantization that are hard to independently validate