Wan 2.2 T2V A14B on 12 GB of VRAM
Not at full quality. The 16-bit and 8-bit files are too big for 12 GB; the largest that fits is Q3_K_M (7.2 GB). Wan 2.2 splits the work between two 14B models: one for the early, noisy steps and one for the late ones. Only one needs to be in VRAM at a time, but both must fit in system RAM.
Q3_K_M (7.2 GB) fits with only 0.5 GB to spare (about 11.5 GB of 12 GB). Quality: noticeable loss of detail. For better quality, Q4_K_M (9.7 GB) also runs, with about 2.0 GB spilling into system RAM - a little slower, still practical.
Also try the FP8 file (14.3 GB) on NVIDIA GPUs. It does not fit entirely, but a current ComfyUI streams the overflow from RAM and native 8-bit files often beat a small GGUF that does fit. Dynamic VRAM →
Using Kijai's WanVideoWrapper instead of the native nodes? Its WanVideo BlockSwap node keeps part of the model in system RAM. To run the FP8 file (14.3 GB) on 12 GB, start with blocks_to_swap ≈ 20 (of 40). Each block is about 0.4 GB; raise the number if you still run out of memory, lower it for speed. estimate WanVideoWrapper →
01Other amounts of VRAM
02Every file, on 12 GB
| File | Size | Needed | On this card | Quality | Download |
|---|---|---|---|---|---|
| 16-bit SAFETENSORS · Comfy-Org | 28.6 GB | 32.9 GB | 20.9 GB too big | the original weights | Hugging Face → |
| Q8_0 GGUF · QuantStack | 15.4 GB | 19.7 GB | 7.7 GB too big | practically identical to the original | Hugging Face → |
| FP8 SAFETENSORS · Comfy-Org | 14.3 GB | 18.6 GB | 6.6 GB too big | practically identical to the original | Hugging Face → |
| Q6_K GGUF · QuantStack | 12.0 GB | 16.3 GB | 4.3 GB too big | very close to the original | Hugging Face → |
| Q5_K_M GGUF · QuantStack | 10.8 GB | 15.1 GB | 3.1 GB too big | close; small differences in fine detail | Hugging Face → |
| Q4_K_M GGUF · QuantStack | 9.7 GB | 14.0 GB | spills 2.0 GB | good; some loss in fine detail and text | Hugging Face → |
| Q3_K_M ← GGUF · QuantStack | 7.2 GB | 11.5 GB | fits · 0.5 GB spare | noticeable loss of detail | Hugging Face → |
| Q2_K GGUF · QuantStack | 5.3 GB | 9.6 GB | fits · 2.4 GB spare | heavy loss; a last resort | Hugging Face → |
“Needed” adds 3.5 GB of working memory for Wan 2.2 T2V and a 0.8 GB reserve. Sizes 2026-10-01. Wan 2.2 A14B uses two files of this size (high-noise and low-noise); only one sits in VRAM at a time, both must fit in system RAM.
03What to download
| Part | File | Folder | Size | |
|---|---|---|---|---|
| Model | Wan2.2-T2V-A14B-HighNoise-Q3_K_M.gguf high-noise model (early steps) Q3_K_M · QuantStack/Wan2.2-T2V-A14B-GGUF | models/unet | 7.2 GB | Download → |
| Model | Wan2.2-T2V-A14B-LowNoise-Q3_K_M.gguf low-noise model (late steps) Q3_K_M · QuantStack/Wan2.2-T2V-A14B-GGUF | models/unet | 7.2 GB | Download → |
| Text encoder | umt5_xxl_fp8_e4m3fn_scaled.safetensors UMT5-XXL FP8 (scaled) · Comfy-Org/Wan_2.1_ComfyUI_repackaged | models/text_encoders | 6.7 GB | Download → |
| VAE | wan_2.1_vae.safetensors Wan 2.1 VAE · Comfy-Org/Wan_2.2_ComfyUI_Repackaged | models/vae | 0.3 GB | Download → |
| Total download · keep about the same free on disk | 21.3 GB | |||
Alternatives: UMT5-XXL FP16 (11.4 GB). Sizes from Hugging Face, 2026-10-01.
Free ComfyUI workflow, tested on my RTX 5060 Ti: FP8 workflow (.json) · GGUF Q5_K_M workflow (.json). Drag the file into ComfyUI. All workflows →