Ideogram 4 on 20 GB of VRAM
Not at full quality. The 16-bit and 8-bit files are too big for 20 GB; the largest that fits is Q5_1 (7.3 GB). Ideogram's open-weights model (June 2026, non-commercial). It uses two transformers at once - a main one and an "unconditional" one - so it needs about twice the memory of one file.
Take Q5_1 (7.3 GB (×2, both loaded)). In use it comes to about 17.0 GB, with 3.0 GB left over on 20 GB. Quality: close; small differences in fine detail. For better quality, FP8 (9.3 GB) also runs, with about 0.9 GB spilling into system RAM - a little slower, still practical.
Also try the FP8 file (9.3 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 →
01What to download
| Part | File | Folder | Size | |
|---|---|---|---|---|
| Model | ideogram4-transformer-q5_1.gguf main model Q5_1 · molbal/ideogram-4-gguf | models/unet | 7.3 GB | Download → |
| Model | ideogram4-unconditional_transformer-q5_1.gguf unconditional model, loaded together with the main one Q5_1 · molbal/ideogram-4-gguf | models/unet | 7.3 GB | Download → |
| Text encoder | qwen3vl_8b_fp8_scaled.safetensors Qwen3-VL 8B FP8 · Comfy-Org/Ideogram-4 | models/text_encoders | 10.6 GB | Download → |
| VAE | flux2-vae.safetensors FLUX.2 VAE · Comfy-Org/Ideogram-4 | models/vae | 0.3 GB | Download → |
| Total download · keep about the same free on disk | 25.6 GB | |||
Alternatives: Qwen3-VL 8B NVFP4 (6.3 GB). FP4 files are listed only as alternatives unless the GPU is an RTX 50 (Blackwell). Sizes from Hugging Face, 2026-10-01.
02Every file, on 20 GB
| File | Size | Needed | On this card | Quality | Download |
|---|---|---|---|---|---|
| Q8_0 GGUF · molbal | 10.1 GB ×2 | 22.6 GB | 2.6 GB too big | practically identical to the original | Hugging Face → |
| FP8 SAFETENSORS · Comfy-Org | 9.3 GB ×2 | 20.9 GB | spills 0.9 GB | practically identical to the original | Hugging Face → |
| INT8 SAFETENSORS · Comfy-Org | 9.6 GB ×2 | 21.5 GB | spills 1.5 GB | practically identical to the original | Hugging Face → |
| Q5_1 ← GGUF · molbal | 7.3 GB ×2 | 17.0 GB | fits · 3.0 GB spare | close; small differences in fine detail | Hugging Face → |
| Q4_1 GGUF · molbal | 6.2 GB ×2 | 14.7 GB | fits · 5.3 GB spare | good; some loss in fine detail and text | Hugging Face → |
“Needed” adds 1.5 GB of working memory for Ideogram 4 and a 0.8 GB reserve. Sizes 2026-10-01. Needs two transformer files loaded together (main + unconditional). GGUF files need the molbal fork of ComfyUI-GGUF. No 16-bit weights were released.