Pony Rating (v6)

Pony v6 prompts open with a small incantation: score_9, score_8_up, score_7_up for quality, a source tag for style, and a rating tag for the content dial. rating_safe keeps it SFW, rating_explicit opens the door, and rating_questionable is the middle ground. It's three words you'll type a thousand times, and "questionable" is exactly the word you'll misspell at 2am.

Pony Rating (v6) generates that tag for you. Pick the rating from a dropdown, get rating_safe, - comma and trailing space included, because the model was trained on the tag list as a whole string and that's how the tags live in a prompt.

How it works

The node is deliberately dumb. One combo input, rating, with the v6 vocabulary: none, safe, questionable, explicit. Choose none and you get an empty string - useful when you want the node in the graph but no rating injected. Choose anything else and the output rating_text is literally rating_{rating}, .

Why the comma and space? Because the quality-tag system is a quirk of how Pony was trained. The tags were fused into one string in the training data, so the whole sequence reads as a quality marker - which is why the score incantation has to appear verbatim. These Sage Pony nodes aren't inventing a system; they're generating the exact fragments the model expects, punctuation included.

What you wire it into

The rating_text output is a STRING, designed to be joined into your prompt with the pack's Join Text / Triple Join Text nodes, or combined with Sage_PonyScore and Sage_PonySource to build the full Pony prefix before it hits a CLIP text encode. There's also a Pony Prefix node in the same family that assembles score + source + rating in one shot if you want the whole header at once.

One genuine tip from the community: put rating_explicit in the positive prompt and steer content in the negative too - Pony users routinely push rating_safe or rating_questionable into the negative to keep SFW images from drifting. The dropdown makes it painless to flip.

Installation

Sage Utils is one pack - ComfyUI Manager, search "Sage Utils", or:

cd ComfyUI/custom_nodes
git clone https://github.com/arcum42/ComfyUI_SageUtils
cd ComfyUI_SageUtils
pip install -r requirements.txt

Restart ComfyUI. Only dependency is dynamicprompts; nothing to download.

Keep in mind

These rating tags are Pony-specific. On Illustrious, SDXL base, or anything newer they do nothing - the KB's Pony notes are explicit that score and rating tags don't transfer across architectures. And if you're generating for Pony v7, use the v7 sibling node instead: the vocabulary there changed (general / sensitive / explicit), and mixing the two vocabularies is a classic way to confuse yourself.