> ## Documentation Index
> Fetch the complete documentation index at: https://justme-8834e675-codex-docs-0-4-44.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart

> Up and running in under 60 seconds.

## Prerequisites

* Linux (Ubuntu 22.04+)
* NVIDIA GPU with CUDA 13 drivers
* [Docker Engine](https://docs.docker.com/engine/install/)
* [NVIDIA Container Toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/)

## Install

```bash theme={null}
curl -fsSL https://ainode.dev/install | bash
```

The installer:

1. Pulls `ghcr.io/getainode/ainode:latest` (\~18 GB, one-time)
2. Writes `/etc/systemd/system/ainode.service`
3. Installs `/usr/local/bin/ainode` host wrapper
4. Starts the service

## Open the UI

```
http://localhost:3000
```

First visit runs onboarding — pick a model and AINode starts loading it in the background. The UI is available immediately; the chat activates once the model finishes loading.

## API

```bash theme={null}
curl http://localhost:8000/v1/models
```

The API is OpenAI-compatible. Point any OpenAI client at `http://localhost:8000/v1`.

```python theme={null}
from openai import OpenAI

client = OpenAI(base_url="http://localhost:8000/v1", api_key="not-needed")
resp = client.chat.completions.create(
    model="Qwen/Qwen2.5-7B-Instruct",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)
```

## Upgrade

```bash theme={null}
ainode update
```

Pulls `:latest` and restarts the service. Your models, config, and training runs are never touched.