Can the RX 9070 XT 16 GB run SCAIL-2 (character animation)?
Download Q3_K_M (9.1 GB). With the model's working memory it needs about 14.4 GB, leaving 1.6 GB spare on 16 GB. Quality: noticeable loss of detail. For better quality, Q4_K_M (11.5 GB) also runs, with about 0.8 GB spilling into system RAM — a little slower, still practical.
01What to download
| Part | File | Folder | Size | |
|---|---|---|---|---|
| Model | wan2.1_14B_SCAIL_2-Q3_K_M.gguf Q3_K_M · vantagewithai/SCAIL-2-GGUF-ComfyUI | models/unet | 9.1 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 | Wan2_1_VAE_bf16.safetensors Wan 2.1 VAE BF16 (Kijai) · Kijai/WanVideo_comfy | models/vae | 0.3 GB | Download → |
| Also needed | clip_vision_h.safetensors CLIP Vision H · Comfy-Org/Wan_2.1_ComfyUI_repackaged | models/clip_vision | 1.3 GB | Download → |
| Total download · keep about the same free on disk | 17.4 GB | |||
Alternatives: UMT5-XXL FP16 (11.4 GB). The 16-bit text encoder is a little more faithful; the smaller one is chosen here because it loads faster and needs less RAM. Only for some workflows: SAM 3.1 multiplex FP16 (segmentation) (1.7 GB, character-replacement workflow (masking)). Sizes read from Hugging Face (2026-09-25). “Download” links start the file directly; the file name links are the same files the official ComfyUI workflows use.
System RAM: 32 GB or more. ComfyUI keeps the model file, the text encoder and the VAE in system RAM and moves them to the GPU as needed. With this set of files that is about 17.4 GB, plus roughly 6 GB for Windows, ComfyUI and a browser: 23.4 GB in total. With less RAM it still runs, but Windows starts swapping to disk and loading gets very slow. calculated
02Every SCAIL-2 file on 16 GB
| File | Size | Needed | On this card | Quality | Download |
|---|---|---|---|---|---|
| 16-bit SAFETENSORS · Comfy-Org | 32.8 GB | 38.1 GB | 22.1 GB too big | the original weights | Hugging Face → |
| FP8 SAFETENSORS · Comfy-Org | 17.7 GB | 23.0 GB | 7.0 GB too big | practically identical to the original | Hugging Face → |
| Q8_0 GGUF · vantagewithai | 18.1 GB | 23.4 GB | 7.4 GB too big | practically identical to the original | Hugging Face → |
| INT8 SAFETENSORS · Comfy-Org | 16.7 GB | 22.0 GB | 6.0 GB too big | practically identical to the original | Hugging Face → |
| Q6_K GGUF · vantagewithai | 14.2 GB | 19.5 GB | 3.5 GB too big | very close to the original | Hugging Face → |
| Q5_K_M GGUF · vantagewithai | 13.0 GB | 18.3 GB | 2.3 GB too big | close; small differences in fine detail | Hugging Face → |
| Q4_K_M GGUF · vantagewithai | 11.5 GB | 16.8 GB | spills 0.8 GB | good; some loss in fine detail and text | Hugging Face → |
| Q3_K_M ← GGUF · vantagewithai | 9.1 GB | 14.4 GB | fits · 1.6 GB spare | noticeable loss of detail | Hugging Face → |
| Q2_K GGUF · vantagewithai | 7.3 GB | 12.6 GB | fits · 3.4 GB spare | heavy loss; a last resort | Hugging Face → |
Sizes from the Hugging Face file listing, read 2026-09-25. “Needed” = file + 4.5 GB working memory for this model + 0.8 GB kept free for the system.
03The text encoder
UMT5-XXL + CLIP Vision H: 11.4 GB as 16-bit, 6.7 GB as FP8, 3.7 GB as GGUF Q4_K_M (plus CLIP Vision H (1.3 GB) for the input image). ComfyUI encodes the prompt first and can push the encoder out of VRAM before sampling, so it does not have to fit together with the model. On 16 GB the FP8 encoder fits on its own, so prompt encoding stays fast.
04About this GPU
ComfyUI supports AMD GPUs on Windows officially since v0.7.0 (January 2026), through ROCm. Memory works the same as on NVIDIA, so the fit verdicts apply. Some custom nodes are NVIDIA-only. RX 9000 (RDNA 4) GPUs have FP8 hardware, and ComfyUI computes FP8 natively on them with PyTorch 2.7+ and ROCm 6.4+.
RX 9070 XT 16 GB: 16 GB GDDR6 · 256-bit · 640 GB/s · RDNA 4. Everything that runs on the RX 9070 XT 16 GB →
05What more VRAM would change
With 24 GB you could run SCAIL-2 at 8-bit or better (FP8, 17.7 GB) with no offloading: for example on the RTX 4090 24 GB, RTX 3090 24 GB, RX 7900 XTX 24 GB.
06Measured and reported results
Nobody has sent measured numbers for this pair yet. If you run SCAIL-2 on a RX 9070 XT 16 GB, send your time per image and peak VRAM and it will appear here, credited.
07Questions
How much VRAM does SCAIL-2 need?
Around 12.6 GB with the smallest file (Q2_K) and 23.0 GB with an 8-bit file (FP8), counting working memory and a small system reserve. The full 16-bit file needs about 38.1 GB.
Which SCAIL-2 file should I download for the RX 9070 XT 16 GB?
Q3_K_M (9.1 GB) from vantagewithai/SCAIL-2-GGUF-ComfyUI. Load it with Unet Loader (GGUF) from the ComfyUI-GGUF node pack.
Is FP8 faster than GGUF on the RX 9070 XT 16 GB?
It can be. This GPU has FP8 hardware, and ComfyUI computes FP8 natively for files made for it (Comfy-Org's fp8_scaled files), or for any FP8 file with the --fast fp8_matrix_mult option. GGUF files are unpacked on the fly, which costs some speed.
How much do I need to download for SCAIL-2 on the RX 9070 XT 16 GB?
About 17.4 GB for the model file, text encoder and VAE listed on this page, and the same again free on disk. With a 16 GB card, 32 GB of system RAM or more is recommended.