Documentation

AgentMesh docs

AgentMesh runs open-weight AI models on computers contributed by independent providers, behind an OpenAI-compatible API. These guides cover using it, hosting it, building on it, and running your own network.

Getting started

What AgentMesh is, who it is for, accounts and roles, and the live links.

Customer guide

Deploy a model, call it with an API key, and understand billing.

Provider guide

Install the worker, enroll machines, manage them, and get paid.

API reference

Authentication, models, deployments, inference, streaming, keys, errors and usage.

Operators & self-hosting

Run your own AgentMesh: configuration, deployment, catalog, pricing, payments and recovery.

AI deployment agent

Coming soon. Gemma 4 on an internal Ollama machine: setup, limits, offline behaviour and tests.

Security & limitations

Trust model, data visibility, controls, residual risks and what is not verified.

Quick start

1Deploy a modelSign up as a customer, pick a model, deploy, and call it with an API key.
2Host modelsSign up as a provider, create an enrollment token, and install the worker.
3IntegrateUse any OpenAI client against your deployment’s /v1 endpoint.

Live links

Dashboard: agent-mesh-ten.vercel.app
Become a provider: agent-mesh-ten.vercel.app/register?role=provider
API base URL: agentmesh-riuz.onrender.com/api/v1
Worker releases: github.com/meshach5667/AgentMeshh/releases
Source code: github.com/meshach5667/AgentMeshh

Your requests run on a computer owned by an independent provider

The provider can see the model, its memory, and every prompt and response this deployment handles. Container isolation protects the provider's machine; it does not hide your data from the provider. Do not send secrets, personal data, or anything confidential. Confidential computing is not available.