Can the RTX 4070 Super 12 GB run HunyuanImage 2.1?
Not at full quality. The 16-bit and 8-bit files are too big for the RTX 4070 Super 12 GB; the largest that fits is Q2_K (7.3 GB).
Q2_K (7.3 GB) fits with only 1.4 GB to spare (about 10.6 GB of 12 GB). Quality: heavy loss; a last resort. For better quality, Q3_K_M (9.8 GB) also runs, with about 1.1 GB spilling into system RAM - a little slower, still practical.
Also try the FP8 file (17.4 GB) on this GPU. 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 →
01What to download
| Part | File | Folder | Size | |
|---|---|---|---|---|
| Model | hunyuanimage2.1-q2_k.gguf Q2_K · calcuis/hunyuanimage-gguf | models/unet | 7.3 GB | Download → |
| Text encoder | qwen_2.5_vl_7b_fp8_scaled.safetensors Qwen2.5-VL 7B FP8 · Comfy-Org/HunyuanImage_2.1_ComfyUI | models/text_encoders | 9.4 GB | Download → |
| Text encoder | byt5_small_glyphxl_fp16.safetensors ByT5 small Glyph-XL · Comfy-Org/HunyuanImage_2.1_ComfyUI | models/text_encoders | 0.4 GB | Download → |
| VAE | hunyuan_image_2.1_vae_fp16.safetensors HunyuanImage 2.1 VAE · Comfy-Org/HunyuanImage_2.1_ComfyUI | models/vae | 0.8 GB | Download → |
| Total download · keep about the same free on disk | 18.0 GB | |||
Alternatives: Qwen2.5-VL 7B BF16 (16.6 GB). Sizes from Hugging Face, 2026-10-01.
02Every HunyuanImage 2.1 file on 12 GB
| File | Size | Needed | On this card | Quality | Download |
|---|---|---|---|---|---|
| 16-bit SAFETENSORS · Comfy-Org | 34.9 GB | 38.2 GB | 26.2 GB too big | the original weights | Hugging Face → |
| FP8 SAFETENSORS · Comfy-Org | 17.4 GB | 20.7 GB | 8.7 GB too big | practically identical to the original | Hugging Face → |
| Q8_0 GGUF · calcuis | 19.8 GB | 23.1 GB | 11.1 GB too big | practically identical to the original | Hugging Face → |
| Q6_K GGUF · calcuis | 15.7 GB | 19.0 GB | 7.0 GB too big | very close to the original | Hugging Face → |
| Q5_K_M GGUF · calcuis | 13.4 GB | 16.7 GB | 4.7 GB too big | close; small differences in fine detail | Hugging Face → |
| Q4_K_M GGUF · calcuis | 11.3 GB | 14.6 GB | 2.6 GB too big | good; some loss in fine detail and text | Hugging Face → |
| Q3_K_M GGUF · calcuis | 9.8 GB | 13.1 GB | spills 1.1 GB | noticeable loss of detail | Hugging Face → |
| Q2_K ← GGUF · calcuis | 7.3 GB | 10.6 GB | fits · 1.4 GB spare | heavy loss; a last resort | Hugging Face → |
“Needed” adds 2.5 GB of working memory for HunyuanImage 2.1 and a 0.8 GB reserve. Sizes 2026-10-01.
03Measured and reported results
Nobody has sent measured numbers for this pair yet. If you run HunyuanImage 2.1 on a RTX 4070 Super 12 GB, send your time per image and peak VRAM and it will appear here, credited.
04With more VRAM
With 24 GB you could run HunyuanImage 2.1 at 8-bit or better (FP8, 17.4 GB) with no offloading: for example on the RTX 4090 24 GB, RTX 3090 24 GB, RX 7900 XTX 24 GB.
What changes from the RTX 4070 Super 12 GB to the RTX 5070 Ti 16 GB → · What changes from the RTX 4070 Super 12 GB to the RTX 3090 24 GB →
05About the RTX 4070 Super 12 GB
The RTX 4070 Super 12 GB is an Ada Lovelace GPU with hardware FP8, so ComfyUI can compute Comfy-Org's FP8 files natively: small and fast. (Plain FP8 files use FP8 maths with the --fast fp8_matrix_mult option.)
12 GB GDDR6X · 192-bit · 504 GB/s · Ada Lovelace. Everything that runs on it →
06Questions
How much VRAM does HunyuanImage 2.1 need?
Around 10.6 GB with the smallest file (Q2_K) and 20.7 GB with an 8-bit file (FP8). The full 16-bit file needs about 38.2 GB.
Which HunyuanImage 2.1 file should I download for the RTX 4070 Super 12 GB?
Q2_K (7.3 GB) from calcuis/hunyuanimage-gguf. Load it with Unet Loader (GGUF) from the ComfyUI-GGUF node pack.
FP8 or Q8_0 for HunyuanImage 2.1 on the RTX 4070 Super 12 GB?
FP8 (17.4 GB): the RTX 4070 Super 12 GB computes it natively, while Q8_0 (19.8 GB) is unpacked on the fly.
How much do I need to download for HunyuanImage 2.1 on the RTX 4070 Super 12 GB?
18.0 GB: hunyuanimage2.1-q2_k.gguf, qwen_2.5_vl_7b_fp8_scaled.safetensors, byt5_small_glyphxl_fp16.safetensors, hunyuan_image_2.1_vae_fp16.safetensors.