Tutorial

A first-hour path through Electros LaunchPad: confirm health, launch a VM, open a shell, run a small model, and call the OpenAI-compatible gateway.

1. Confirm the shell is healthy

1. Open Electros LaunchPad.
2. Look at the sidebar footer — it should say All systems healthy.
3. Open Installers and confirm llmfit, llama.cpp, and MLX show Ready.

Details: Getting started, System tools.

2. Launch a VM from Images

1. Go to Compute → Images.
2. Pick a light image (Alpine or Ubuntu Server).
3. Choose a version; note Minimum disk.
4. Click Launch.

Orange Launch uses defaults. Use Configure if you need a custom name or more RAM.

3. Open the instance shell

1. Open Compute → Instances.
2. Wait until STATE is Running.
3. Click the shell icon (or the name) → stay on Shells.
4. Run a simple command at the prompt.

On Alpine, prefer doas if sudo is missing.

4. Download and load a model

1. AI → Models → Top picks.
2. Download a small chat model.
3. AI → Downloaded → Load.
4. Open Running models, click the model, send a prompt in Try this model.

A short prompt here proves the runner and accelerator before you wire external clients.

5. Create a gateway key

1. Open AI → Credentials.
2. Create key under API keys; copy the sk-elp-… secret.
3. From this Mac:

curl http://127.0.0.1:11434/v1/models \
  -H "Authorization: Bearer sk-elp-YOUR_KEY"

Use the From a VM HTTPS URL when the client runs inside a LaunchPad instance.

6. Optional: deploy Open WebUI

1. Services → Catalogue → Open WebUI → Deploy.
2. Point Open WebUI at the gateway URL + your key.
3. Chat against the model you loaded.

Where to go next

• Compositions — multi-node stacks
• Credentials & runner — LAN HTTPS and cloud providers
• CLI create instance — same VMs from elp