# AMIs and machine images

Source: https://docs.quake.ai/docs/compute/concepts/amis
Markdown: https://docs.quake.ai/docs/compute/concepts/amis.md

---

# AMIs and machine images

An AWS **AMI** (Amazon Machine Image) is a boot template for EC2. On Quake AI the same role is an **[image](/docs/compute/concepts/images)** in the Images service (OpenStack Glance): a disk template with metadata (minimum disk, OS family) that Nova uses when it creates an **instance**.

**Instance snapshots** can also become bootable images when you capture a running disk for clone or golden-image workflows.

## Mapping image vocabulary

| Elsewhere | On Quake AI |
|---|---|
| AMI | [Image](/docs/compute/concepts/images) |
| Azure managed image / gallery image | [Image](/docs/compute/concepts/images) |
| GCP machine image | [Image](/docs/compute/concepts/images) |
| Custom image upload | Glance image import (see [create an image](/docs/compute/how-to/create-image)) |
| Golden image pipeline | Packer publish to Glance (see [build a golden image](/docs/compute/how-to/build-golden-image-packer)) |

Images are project-scoped unless shared. Pick an image that matches your workload OS and boot mode requirements before you select a flavor.

## What to read next

- [Images](/docs/compute/concepts/images): catalog, custom images, and visibility
- [Flavors](/docs/compute/concepts/flavors): CPU, RAM, and root disk pairing with images
- [Instance snapshots](/docs/compute/concepts/snapshots): capture instance disk state into an image