OG Labs

0G Labs operates 0G (also styled ØG), a decentralized infrastructure platform positioned as a trust layer for artificial intelligence. Its stated core capabilities are verifiable compute, sealed private inference and decentralized storage, intended for AI agents, developers, enterprises, researchers, consumers, AI applications, robotics, AI startups, AI data providers, AI marketplaces, machine economies and governments.

The platform is organized as a modular stack. An "AI Private Cloud" layer covers decentralized infrastructure for throughput and security, workload execution through APIs, and developer experience tooling including SDKs and quick-start flows. An "Agentic Alignment" layer covers open and expert AI models, trust-and-safety mechanisms such as reputation, attestations and verifiable inference, agent harnesses providing frameworks, skills and persistent memory, and financial primitives for machine-to-machine activity including payments, stablecoins and agent wallets. An applications layer spans autonomous agents, ecosystem products such as launchpads and trading plus the 0G App suite, network operations, and physical AI covering robots, drones and humanoids.

Shipped products listed by the company include the ØG App, Private Computer and Chain Hub, alongside models 0GM-1.0-35B-A3B, GLM-5, GLM-5.1 and GLM-5.2. The company also publishes research, in several cases co-authored with academic and corporate labs including the National University of Singapore, Peking University, BUPT, Nanyang Technological University, Microsoft Research, Chongqing University and Renmin University of China, on topics such as decentralized large-scale training (DiLoCoX), consensus-protocol bug detection with LLM agents, deep learning job deadlock detection, distributed data-parallel computation optimization, operating-system agents and visual reasoning.

Traction

Company-reported network figures include 309.19 billion total private AI tokens consumed, 29.1 million total transactions, 42.9 million blocks produced, 5.5 million total accounts, 2 AI models built and more than 250 ecosystem partners. The company maintains a builder community with public X and Discord channels and publishes research papers with academic and industry collaborators dated from 2024 through 2026.