# Getting Started

URL: https://docs.hanzo.ai/docs/studio/getting-started

> Run Hanzo Studio locally, submit your first prompt, and scale to GPU.

Hanzo Studio is a visual, node-based AI engine. Start on CPU for prototyping,
then switch to GPU with one API call when you need throughput.

## Run locally

```bash title="Docker"
docker pull ghcr.io/hanzoai/studio:latest
docker run -p 8188:8188 ghcr.io/hanzoai/studio:latest \
  python main.py --listen 0.0.0.0 --cpu
# Open http://localhost:8188
```

The editor opens in your browser. Drag nodes onto the canvas, wire them together,
and press **Queue Prompt** to run the graph.

## Authenticate

Studio authenticates against Hanzo IAM. Get an API key at
[console.hanzo.ai](https://console.hanzo.ai) and pass it as a bearer token:

```bash
curl -H "Authorization: Bearer $HANZO_API_KEY" \
  https://studio.hanzo.ai/v1/studio/prompts \
  -d '{"workflow": { /* ... */ }}'
```

> One credential works across every Hanzo service. The same `$HANZO_API_KEY`
> authenticates Studio, Chat, the gateway, and the rest.

## Scale to GPU

Each organization has a compute profile. Switch it at any time — the coordinator
routes GPU-heavy prompts to dedicated workers.

| Profile     | GPU  | VRAM  | Best for                |
| ----------- | ---- | ----- | ----------------------- |
| `cpu`       | —    | —     | Prototyping             |
| `gpu-basic` | T4   | 16 GB | Standard diffusion      |
| `gpu-pro`   | A100 | 80 GB | Video, 3D, large models |
| `gpu-max`   | H100 | 80 GB | Maximum throughput      |

## Next steps

- Browse the [API reference](https://docs.hanzo.ai/docs/openapi/engine) — generated from the OpenAPI spec.
- Every ComfyUI custom node works unmodified; the ecosystem manual is mirrored
  under [Open Source → ComfyUI](https://docs.hanzo.ai/docs/projects).