Skip to main content
nemu routes your applications to the model providers you already pay for, and gives you an assistant and a real machine built on the same plumbing. You bring your own provider keys. nemu holds them, picks the model, enforces your limits, records what happened and forwards the request. Your code only ever sees a nemu key.

Quickstart

A provider, a model, a key and a working request in about five minutes.

How nemu is put together

The console, the gateway and chat, and how the three fit together.

Use it from your own code

How the gateway works

Base URLs, every endpoint, how a model name resolves, and what an error means.

Claude Code

Point the CLI at nemu and answer it with any model you have connected.

Codex

Add nemu as a custom provider over the OpenAI Responses API.

API reference

The management API for everything the console does.

The console

Providers and models

Connect a provider, expose a model under a name of your choosing.

Workflows

Route a request through logic before it reaches a model.

Chat and the Computer

Chat

The assistant, its tools, and the work it does in the background.

Computer

A real machine attached to a chat, billed for the minutes it runs.

Plans

Plans and pricing

What each plan costs and exactly what it allows.

Credits

The one balance everything metered draws from, and what does not.

Packages

@nemu-ai/pdf

Build PDFs from a document model, with tables, charts and formulas.

OAuth 2.0

Authorize an application against the management API.