---
title: "Pony Diffusion V6 XL Low-VRAM & 8GB/12GB Settings in ComfyUI"
description: "How to run Pony Diffusion V6 XL on 8 GB and 12 GB Windows GPUs in ComfyUI using FP8, GGUF quantization, weight offloading, and tiled VAE decoding."
canonical: https://comfy.ink/models/pony-diffusion-v6-xl-low-vram-settings
updated: 2026-10-17
---

# Pony Diffusion V6 XL Low-VRAM Settings for ComfyUI

Step-by-step memory tuning for Pony Diffusion V6 XL (8 GB+) so 8 GB and 12 GB RTX cards render cleanly without CUDA Out of Memory errors.

## Facts (2026-10-17)

- **Target Model:** Pony Diffusion V6 XL
- **Baseline VRAM:** 8 GB+
- **Quantization Options:** FP16 safetensors
- **VAE Strategy:** VAE Decode (Tiled) 512px tiles

## Fitting Pony Diffusion V6 XL into 8 GB–12 GB VRAM

Standard full-precision weights for [Pony Diffusion V6 XL](https://comfy.ink/models/pony-diffusion-v6-xl) target 8 GB+. On 8 GB cards like the RTX 3060 Ti, RTX 4060, or laptop GPUs, loading uncompressed weights alongside the text encoder can trigger a CUDA Out of Memory error.

Switching to quantized weights (FP16 safetensors) via [ComfyUI-GGUF](https://comfy.ink/nodes/gguf-loader-nodes) or native FP8 checkpoints cuts the model footprint roughly in half while keeping prompt adherence intact.

## Checklist to prevent OOM spikes at the end of sampling

1. Replace standard `VAE Decode` with [VAE Decode (Tiled)](https://comfy.ink/nodes/vae-decode-encode) so the final latent-to-pixel conversion runs in 512×512 blocks.
2. Use `t5xxl_fp8_e4m3fn.safetensors` or a Q4 GGUF text encoder when running dual/triple CLIP architectures.
3. Keep batch size at 1 and use Queue Count in [ComfyUI Studio](https://comfy.ink/studio) when generating multiple variations. Grab the latest Windows build from [/download](https://comfy.ink/download).

## FAQ

### Why does Pony Diffusion V6 XL crash at 95% or 100% in ComfyUI?

A crash right as sampling finishes is almost always the VAE decode step spiking VRAM. Swap in the `VAE Decode (Tiled)` node to solve it immediately.

### Does --lowvram slow down Pony Diffusion V6 XL on Windows?

ComfyUI automatically enables smart weight offloading when VRAM is tight; only pass `--lowvram` if a heavy custom node bypasses ComfyUI's native memory manager.

## Related

- [Pony Diffusion V6 XL in ComfyUI](https://comfy.ink/models/pony-diffusion-v6-xl)
- [ComfyUI Models & Checkpoints](https://comfy.ink/models)
- [How to Fix CUDA Out of Memory (OOM) in ComfyUI](https://comfy.ink/howto/fix-cuda-out-of-memory-comfyui)
- [Running ComfyUI on 8 GB VRAM GPUs (RTX 3060 / 4060)](https://comfy.ink/guides/comfyui-low-vram-8gb-gpu-guide)
- [Flux.1 Dev in ComfyUI](https://comfy.ink/models/flux-1-dev)

## Download

Free desktop app: [Download Comfy](https://comfy.ink/download)

Markdown mirror: https://comfy.ink/models/pony-diffusion-v6-xl-low-vram-settings.md