GitHub Repository

A decentralized peer-to-peer network for sharing and accessing Large Language Models (LLMs) through a standardized OpenAI-compatible API.

6 starsRust

A peer-2-peer network for sharing and accessing LLMs through OpenAI API

Name: A peer-2-peer network for sharing and accessing LLMs through OpenAI API
Availability: InStock
Author: modinfo

by modinfo·Feb 25, 2026·1 point·1 comment

Visit Project View on HN

AI Analysis

●MidBold BetBig Brain

Distributed LLM inference over P2P instead of centralized APIs, but early-stage and unproven.

Strengths

•Architecture is thoughtful: Kademlia DHT for discovery, libp2p for transport, OpenAI API compatibility for drop-in use.
•Rate limiting and schedule-based access control built in—considers provider safety.

Weaknesses

•No working demo, live network, or user reports; unclear if routing actually works across real latency/NAT boundaries.
•Competes with Ollama (local), Together AI, and modal (distributed inference)—differentiator is unclear beyond 'peer-to-peer ideology.'
•Zero evidence of adoption or real-world testing; feels like a well-architected proof of concept, not a usable product yet.

Post Description

Conduit allows users to securely serve their own local models (or proxy) and access models from others globally, creating a distributed and resilient compute network.