TLDR Tech / Crypto
Dolphin Network Targets Idle GPU Inference Market
Dolphin Network routes AI inference jobs across idle consumer GPUs in a peer-to-pool model, directing 100% of protocol revenue toward open-market $POD buybacks to offset emissions. The pricing structure captures a $0.20/M token margin on Qwen 3.6 35B inference at $0.70/M versus OpenRouter's $1.00, positioning Dolphin 30% below the cheapest centralized provider while paying nodes $0.50/M. Token design draws from veCRV boosts, xSUSHI/yCRV auto-compounding, stAAVE cooldowns, and vlCVX bribe markets, with a 20% liquid-claim fee routing to the xPOD vault and a bond-duration multiplier reaching 2x beyond six months. A forthcoming paper titled "Encrypted Live-Weight Proofs for Decentralized Inference" addresses model-correctness verification across all hardware types, extending coverage beyond TEE-only NVIDIA cards.
Read the full story