Check my GPU →

Wan 2.1 VACE 14B on 6 GB of VRAM

Not entirely. Even Q4_K_M is 10.4 GB over 6 GB, so ComfyUI keeps part of it in system RAM. The all-in-one Wan for control and editing: reference-to-video, pose and depth control, inpainting and extending clips.

Video model · 17.3BAny 6 GB GPUData 2026-10-01
Offload
Not entirely in VRAM, but it can still run.

The smallest sensible file, Q4_K_M (11.6 GB), needs about 16.4 GB - 10.4 GB more than 6 GB GPU has. ComfyUI can still run it by streaming part of the model from system RAM. How much slower that is depends on your ComfyUI version, the file format and the PCIe link - see the note below.

Best fileQ4_K_M
File size11.6 GB
VRAM needed~16.4 GB
System RAM32 GB+
Memory map · 6 GB GPU16.4 GB needed · 10.4 GB over 6 GB
08 GB16 GB
Weights Q4_K_M · 11.6 GBWorking memory · 4.0 GBReserve · 0.8 GBSpills to system RAM · 10.4 GB
calculated How this is counted.

01Other amounts of VRAM

8 GB Offload only10 GB Offload only12 GB Offload only16 GB Tight20 GB Runs24 GB Runs well32 GB Runs well

02Every file, on 6 GB

FileSizeNeededOn this cardQualityDownload
16-bit
SAFETENSORS · Comfy-Org
34.7 GB39.5 GB33.5 GB too bigthe original weightsHugging Face →
Q8_0
GGUF · QuantStack
18.7 GB23.5 GB17.5 GB too bigpractically identical to the originalHugging Face →
Q6_K
GGUF · QuantStack
14.5 GB19.3 GB13.3 GB too bigvery close to the originalHugging Face →
Q5_K_M
GGUF · QuantStack
13.0 GB17.8 GB11.8 GB too bigclose; small differences in fine detailHugging Face →
Q4_K_M ←
GGUF · QuantStack
11.6 GB16.4 GB10.4 GB too biggood; some loss in fine detail and textHugging Face →
Q3_K_S
GGUF · QuantStack
7.8 GB12.6 GB6.6 GB too bignoticeable loss of detailHugging Face →

“Needed” adds 4 GB of working memory for Wan VACE 14B and a 0.8 GB reserve. Sizes 2026-10-01.

03What to download

PartFileFolderSize
ModelWan2.1_14B_VACE-Q4_K_M.gguf
Q4_K_M · QuantStack/Wan2.1_14B_VACE-GGUF
models/unet11.6 GBDownload →
Text encoderumt5_xxl_fp8_e4m3fn_scaled.safetensors
UMT5-XXL FP8 (scaled) · Comfy-Org/Wan_2.1_ComfyUI_repackaged
models/text_encoders6.7 GBDownload →
VAEwan_2.1_vae.safetensors
Wan 2.1 VAE · Comfy-Org/Wan_2.1_ComfyUI_repackaged
models/vae0.3 GBDownload →
Total download · keep about the same free on disk18.6 GB

Alternatives: UMT5-XXL FP16 (11.4 GB). Sizes from Hugging Face, 2026-10-01.

04The text encoder

Take the smallest UMT5-XXL file (3.7 GB): it is the only one that fits in 6 GB on its own. Sizes: 16-bit 11.4 GB, FP8 6.7 GB, GGUF Q4_K_M 3.7 GB. Why the encoder does not count against VRAM →