AI Models

The per-project list of AI models you want to use -- turn models on, and see exactly which files that selection needs and which you already have.

AI Models is a project view -- a per-project allow-list of the AI models you want to use on a project. It shapes:

  • Which recipes and workflows appear when you open Add a step
  • Which models the platform suggests for new recipes
  • What you still need to install for the models you have chosen
  • A picture of what's available on Comfy Cloud vs your local ComfyUI

Open it from the Actions tab on any shot, character, location, or prop page -- the AI Models chip in the empty state, or the + Add model chip beside any operation heading in the operation picker.

The page follows one simple order: choose the models you want first, then resolve what they need. It never downloads the whole catalogue up front.

Getting set up

The first time through, a short setup card guides the choices that come before picking models, one question at a time:

  • Whether you have a Comfy Cloud account you want to use.
  • Whether you want to run models locally -- already set up, want help setting up, or cloud only.
  • If local: a prompt to connect your models folder (see below).

It only asks what it doesn't already know -- if you have a Comfy Cloud key saved, or your models folder is already connected, it skips straight past those questions and won't ask again. Once you are set up, the card steps aside. Not sure which models to pick? The card's Help me choose button opens the assistant, which reads your project and hardware and recommends a model.

Turning models on

Each model is shown as a card with a thumbnail, its provider's logo, Local | Cloud chips showing where it's available, and a family chip (e.g. flux, sdxl, wan22). Click a card to turn the model on or off for this project -- the green primary border marks the ones that are on. Your selection saves as you go.

When your local models folder is connected, each card also shows a small status chip: On disk, Found nearby, Installing, or an Install button for that one file.

Will it run on your machine?

Once your account knows your graphics card, every model card answers that question up front -- while you are choosing, not after you have downloaded several gigabytes:

Chip What it means
Runs on your card It fits in your VRAM and will run at normal speed.
Slow on your card It only fits by spilling out of VRAM, so expect it to be slow.
Cloud only Too big for your card, but Comfy Cloud can run it. Where a per-run estimate is known, the chip carries the approximate cost.
Too big for your card Too big locally, and you have no cloud connected -- so there is currently no way to run it.

Hover any chip for the reasoning behind it. The chip is deliberately silent in two cases: when your card is not known (a fit we never checked is exactly the claim this is meant to prevent), and for hosted services, where the Local | Cloud chips beside it already say where they run. Your graphics card and VRAM live in your account settings, not in the project -- see Choosing the right model.

Files needed

As soon as you turn on models that need files you don't have locally, this page tells you so and offers to take you to the ComfyUI section of Settings, where getting files actually happens. That is where the Files needed list lives, covering just your current selection -- not the catalogue. For each file it shows:

  • Its name and size, with a running total, so you can confirm you have the disk space before any multi-gigabyte download.
  • Its location (the folder it belongs in) and status:
    • On disk -- already there.
    • Found in <folder> -- you already have this file, just not where the platform expects it (see below).
    • Ready to download -- the platform can fetch it for you; tick it and choose Install selected.
    • Manual install -- no automatic download; the list gives you the link and the exact folder to place it in.

You choose what to install; nothing large downloads without you confirming it.

Files you already have

If you already run ComfyUI, you probably have many of these files under your own names and folders. When the list finds one that isn't where the platform expects, it says Found in <folder> and offers a choice instead of proposing a duplicate download:

  • Move here -- move your file to the platform's standard location.
  • Copy here -- copy it there, leaving your original in place.
  • Use existing file -- reference your file exactly where it sits, changing nothing on disk. This keeps any of your own workflows that point at that file working. It's offered when the file is in the right kind of folder for that model.

Connecting your models folder

Connecting your ComfyUI models folder is a one-time setup that lets the platform see which files you already have and install missing ones. It lives under Settings, in the ComfyUI section, with a short walkthrough of how to find your ComfyUI folder. Once connected, it is remembered across visits, and the AI Models page uses it automatically. This works in Chromium browsers such as Chrome and Edge; elsewhere the Files needed list still shows you the links and folders to install by hand.

Filtering and finding models

The header bar gives you several ways to narrow the list:

  • Search -- filters by name or model ID
  • N selected toggle -- show only the models currently turned on
  • Local toggle -- show only models installed on your local ComfyUI
  • Cloud toggle -- show only models available on Comfy Cloud
  • + Filter dropdown -- filter by operation (Text to image, Lipsync, Voice clone, etc.)

All filters combine as AND with the text search. The + Filter dropdown takes the same operation taxonomy as the workflow picker, so opening AI Models from a specific operation (via + Add model) lands you straight on the matching section. When a section has no models registered yet, its header still appears with a note explaining the gap.

Refreshing local availability

The Refresh button re-probes your local ComfyUI install to update the Local indicators -- which model files and which API-node integrations are present -- so you get an accurate picture without leaving the page.

Learn more