⚡ Quick Answer
A quantized model stores its weights using fewer bits of precision than it was originally trained with. This shrinks the file size and the VRAM needed to run it, at some cost to output quality. GGUF and fp8 are the two quantization formats you'll run into most often in ComfyUI.
Quantization is the general idea; GGUF and fp8 are two specific, different ways of putting that idea into practice.
Where You'll See It
Model download pages label a smaller release as "quantized," or name its specific level directly — "Q4," "Q8," or "fp8" — next to the full-precision original, usually released as bf16 or fp16.
Quick Example
A checkpoint's full bf16 release might be around 24 GB, while its quantized fp8 or GGUF release comes in closer to 8–12 GB — letting it run on a much smaller graphics card.
Frequently Asked Questions
See It In Action
Not sure which version to download?
Our VRAM & Hardware Guide matches models to what your GPU can actually run.
Published: 2026-09-17 · Last updated: 2026-09-17
Join the discussion
Sign in to leave a comment or reply
No comments yet
Be the first to share your thoughts!
