Curated Models, if you just need a working chat or coding model fast. Templates, if you want a specific outcome like text-to-image without assembling anything. Custom Models and Services are for when you need more control.

Yes. Each of those flows has a "Customize" or "Advanced setup" option that hands you off into the same full environment picker Custom Models uses, so you're not locked into the pre-sized default if you want to size the booking yourself.

No. Your selection lives in the page URL at every stage, so refreshing, bookmarking, or sharing the link resumes the same flow. The one exception is secrets, access tokens are held only in your browser tab and are never written into the URL.

Yes, as long as you sign in with the same wallet. Every service you've launched is listed on the Inference landing page tied to your wallet, so you can pick up any of them from another device.

Curated and Custom Models expose an OpenAI-compatible API endpoint you can call directly. Services and Templates instead give you the app's own web UI, opened in a new tab, there's no API call, you just use the interface.

The environment picker tells you outright, if the matching nodes are out of capacity, the panel says so rather than letting you pick an environment that would just fail to launch.