Plans

Quake AI concept: Flavors

• Vultr treats instance sizes as predefined Plans, defined as a particular configuration of vCPU, RAM, SSD, and bandwidth, rather than OpenStack Nova flavors with arbitrary extra_specs or tenant-defined flavor creation. The canonical API exposes this as `GET /v2/plans` and region-specific availability via `GET /v2/regions/{region-id}/availability`.
• Plan pricing is baked into the plan catalog itself. The Vultr Plans response includes `monthly_cost`, `hourly_cost`, preemptible pricing fields, and per-location cost data, which is different from OpenStack's flavor model where the flavor object itself is not a billing construct.
• Plan selection is constrained by availability in a specific region. Vultr explicitly says not all plans are available in all regions, and the region availability endpoint returns only the plans offered there.
• Resizing is asymmetric: updating an instance to a new plan is supported, but downgrading is not supported according to Vultr's resize documentation. The resize flow uses the instance update API and a separate job endpoint, rather than a generic Nova flavor change pattern.