Ideogram 4 VRAM requirements
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.
TypeImage
Parameters9.3B
Fits entirely from15 GB
8-bit or better from21 GB
01The files, and how much VRAM each needs
| File | Size | Needed | Min. VRAM | Quality | Source |
|---|---|---|---|---|---|
| Q8_0 | 10.1 GB ×2 | 22.6 GB | 23 GB | practically identical to the original | molbal/ideogram-4-gguf → |
| FP8 | 9.3 GB ×2 | 20.9 GB | 21 GB | practically identical to the original | Comfy-Org/Ideogram-4 → |
| INT8 | 9.6 GB ×2 | 21.5 GB | 22 GB | practically identical to the original | Comfy-Org/Ideogram-4 → |
| Q5_1 | 7.3 GB ×2 | 17.0 GB | 17 GB | close; small differences in fine detail | molbal/ideogram-4-gguf → |
| Q4_1 | 6.2 GB ×2 | 14.7 GB | 15 GB | good; some loss in fine detail and text | molbal/ideogram-4-gguf → |
“Needed” = file + 1.5 GB working memory + 0.8 GB system reserve. Needs two transformer files loaded together (main + unconditional). GGUF files need the molbal fork of ComfyUI-GGUF. No 16-bit weights were released.
02By amount of VRAM
03Best GPU for Ideogram 4
The cheapest cards (by launch price) that run it well, and every card sorted by memory: best GPU for Ideogram 4 → Planning bigger images or longer clips? Open the calculator →
04By graphics card
| GPU | VRAM | Verdict | Best file | Needed |
|---|---|---|---|---|
| Desktop graphics cards | ||||
| RTX 2060 6 GB | 6 GB | Not practical | Q4_1 | 14.7 GB |
| RTX 3050 6 GB | 6 GB | Not practical | Q4_1 | 14.7 GB |
| RTX 2070 Super 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 2080 Super 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3050 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3060 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3060 Ti 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3070 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3070 Ti 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 4060 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 4060 Ti 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5050 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5060 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5060 Ti 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RX 7600 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RX 9050 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RX 9060 XT 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| Arc B570 10 GB | 10 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3080 10 GB | 10 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 2080 Ti 11 GB | 11 GB | Offload only | Q4_1 | 14.7 GB |
| Arc B580 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 2060 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3060 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3080 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3080 Ti 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 4070 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 4070 Super 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 4070 Ti 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5070 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RX 7700 XT 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RX 9070 GRE 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| Arc A770 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 4060 Ti 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 4070 Ti Super 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 4080 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 4080 Super 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 5060 Ti 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 5070 Ti 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 5080 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RX 7600 XT 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RX 7800 XT 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RX 7900 GRE 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RX 9060 XT 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RX 9070 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RX 9070 XT 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RX 7900 XT 20 GB | 20 GB | Runs | Q5_1 | 17.0 GB |
| Arc Pro B60 24 GB | 24 GB | Runs well | Q8_0 | 22.6 GB |
| RTX 3090 24 GB | 24 GB | Runs well | Q8_0 | 22.6 GB |
| RTX 3090 Ti 24 GB | 24 GB | Runs well | Q8_0 | 22.6 GB |
| RTX 4090 24 GB | 24 GB | Runs well | FP8 | 20.9 GB |
| RX 7900 XTX 24 GB | 24 GB | Runs well | Q8_0 | 22.6 GB |
| Arc Pro B70 32 GB | 32 GB | Runs well | Q8_0 | 22.6 GB |
| RTX 5090 32 GB | 32 GB | Runs well | FP8 | 20.9 GB |
| Laptop GPUs | ||||
| RTX 3050 Laptop 4 GB | 4 GB | Not practical | Q4_1 | 14.7 GB |
| RTX 3050 Ti Laptop 4 GB | 4 GB | Not practical | Q4_1 | 14.7 GB |
| RTX 2060 Laptop 6 GB | 6 GB | Not practical | Q4_1 | 14.7 GB |
| RTX 3050 Laptop 6 GB | 6 GB | Not practical | Q4_1 | 14.7 GB |
| RTX 3060 Laptop 6 GB | 6 GB | Not practical | Q4_1 | 14.7 GB |
| RTX 4050 Laptop 6 GB | 6 GB | Not practical | Q4_1 | 14.7 GB |
| RTX 2070 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 2070 Super Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 2080 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 2080 Super Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3070 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3070 Ti Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3080 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 4060 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 4070 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5050 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5060 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5070 Laptop 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RX 7600M 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RX 7600M XT 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RX 7600S 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RX 7700S 8 GB | 8 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 4080 Laptop 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5070 Laptop 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 5070 Ti Laptop 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RX 7800M 12 GB | 12 GB | Offload only | Q4_1 | 14.7 GB |
| RTX 3080 Laptop 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 3080 Ti Laptop 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 4090 Laptop 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 5080 Laptop 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RX 7900M 16 GB | 16 GB | Runs | Q4_1 | 14.7 GB |
| RTX 5090 Laptop 24 GB | 24 GB | Runs well | FP8 | 20.9 GB |
| Unified memory | ||||
| Radeon 8060S (Strix Halo) 96 GB | 96 GB | Runs well | Q8_0 | 22.6 GB |
| Radeon 8065S (Gorgon Halo) 160 GB | 160 GB | Runs well | Q8_0 | 22.6 GB |
| Apple Silicon Macs (by memory) | ||||
| Mac 16 GB | 12.7 GB | Offload only | Q4_1 | 14.7 GB |
| Mac 18 GB | 14.4 GB | Offload only | Q4_1 | 14.7 GB |
| Mac 24 GB | 19.6 GB | Runs | Q5_1 | 17.0 GB |
| Mac 32 GB | 26.8 GB | Runs well | Q8_0 | 22.6 GB |
| Mac 36 GB | 30.2 GB | Runs well | Q8_0 | 22.6 GB |
| Mac 48 GB | 40.2 GB | Runs well | Q8_0 | 22.6 GB |
| Mac 64 GB | 55.7 GB | Runs well | Q8_0 | 22.6 GB |
| Mac 96 GB | 85 GB | Runs well | Q8_0 | 22.6 GB |
| Mac 128 GB | 115.4 GB | Runs well | Q8_0 | 22.6 GB |
| Mac 192 GB | 175.4 GB | Runs well | Q8_0 | 22.6 GB |
| Mac 256 GB | 236.9 GB | Runs well | Q8_0 | 22.6 GB |
| Mac 512 GB | 498.1 GB | Runs well | Q8_0 | 22.6 GB |
05Text encoder, VAE and other files
| File | Folder | Size | When | |
|---|---|---|---|---|
| Qwen3-VL 8B FP8 qwen3vl_8b_fp8_scaled.safetensors | models/text_encoders | 10.6 GB | text encoder · smaller, recommended | Download → |
| Qwen3-VL 8B NVFP4 qwen3vl_8b_nvfp4.safetensors | models/text_encoders | 6.3 GB | text encoder · alternative | Download → |
| FLUX.2 VAE flux2-vae.safetensors | models/vae | 0.3 GB | required | Download → |
The files the official ComfyUI workflows load next to the model. Sizes read from Hugging Face (2026-10-01).
06Where the files go in ComfyUI
| File | Folder | Loader node |
|---|---|---|
| Diffusion model (.safetensors: 16-bit, FP8, INT8) | ComfyUI/models/diffusion_models | Load Diffusion Model |
| GGUF file (.gguf) | ComfyUI/models/unet | Unet Loader (GGUF) - from the ComfyUI-GGUF node pack |
| Text encoder | ComfyUI/models/text_encoders | Load CLIP / DualCLIPLoader (or the GGUF versions) |
| VAE | ComfyUI/models/vae | Load VAE |
07Measured and reported results
No measured results yet. Send yours.
08Training a LoRA for Ideogram 4
No trainer documentation with a VRAM figure for this model was found yet. What the trainers say for other models →
09Every file tracked for Ideogram 4
| File | Type | Size | Repo |
|---|---|---|---|
| ideogram4_int8_convrot.safetensors (main) | INT8 | 9.58 GB | Comfy-Org/Ideogram-4 → |
| ideogram4_unconditional_int8_convrot.safetensors (unconditional) | INT8 | 9.58 GB | Comfy-Org/Ideogram-4 → |
| ideogram4_unconditional_fp8_scaled.safetensors (unconditional) | FP8 | 9.28 GB | Comfy-Org/Ideogram-4 → |
| ideogram4_fp8_scaled.safetensors (main) | FP8 | 9.28 GB | Comfy-Org/Ideogram-4 → |
| ideogram4_nvfp4_mixed.safetensors (main) | NVFP4 | 5.49 GB | Comfy-Org/Ideogram-4 → |
| ideogram4_unconditional_nvfp4_mixed.safetensors (unconditional) | NVFP4 | 5.49 GB | Comfy-Org/Ideogram-4 → |
| ideogram4-transformer-q8_0.gguf (main) | Q8_0 | 10.14 GB | molbal/ideogram-4-gguf → |
| ideogram4-unconditional_transformer-q8_0.gguf (unconditional) | Q8_0 | 10.14 GB | molbal/ideogram-4-gguf → |
| ideogram4-transformer-q5_1.gguf (main) | Q5_1 | 7.33 GB | molbal/ideogram-4-gguf → |
| ideogram4-unconditional_transformer-q5_1.gguf (unconditional) | Q5_1 | 7.33 GB | molbal/ideogram-4-gguf → |
| ideogram4-transformer-q5_0.gguf (main) | Q5_0 | 6.77 GB | molbal/ideogram-4-gguf → |
| ideogram4-unconditional_transformer-q5_0.gguf (unconditional) | Q5_0 | 6.77 GB | molbal/ideogram-4-gguf → |
| ideogram4-transformer-q4_1.gguf (main) | Q4_1 | 6.21 GB | molbal/ideogram-4-gguf → |
| ideogram4-unconditional_transformer-q4_1.gguf (unconditional) | Q4_1 | 6.21 GB | molbal/ideogram-4-gguf → |
| ideogram4-transformer-q4_0.gguf (main) | Q4_0 | 5.64 GB | molbal/ideogram-4-gguf → |
| ideogram4-unconditional_transformer-q4_0.gguf (unconditional) | Q4_0 | 5.64 GB | molbal/ideogram-4-gguf → |
9.3B single-stream DiT, open weights from Ideogram (June 2026). Uses dual-branch CFG: workflows load BOTH the main and the 'unconditional' transformer (two ~9.3 GB fp8 files), so real VRAM need is roughly double one file unless ComfyUI offloads between passes. VAE: flux2-vae.safetensors (FLUX.2 VAE).