by Ideogram
Ideogram 4 is a 9.3B single-stream DiT text-to-image foundation model trained from scratch, with best-in-class in-image text rendering for signage, logos, and captions. Uses a 34-layer unified transformer over concatenated text and image tokens, with Qwen3-VL-8B as the text encoder. Supports any multiple-of-16 resolution up to 2048x2048.
VRAM requirements, GPU fit, and setup notes for Ideogram 4, including 8GB/12GB fit guidance where relevant. Recommended runtimes: ComfyUI and Diffusers support. Best download size: ~5.8 GB at Q4_K_M.
Your hardware
Detecting...
Measured quality metrics for Ideogram 4 outputs.
How often humans prefer this model's output (0-100%)
Visual quality and composition rating (5-9 scale)
Text-image alignment accuracy (higher is better)
Compare which GPUs can run Ideogram 4 at different precisions. FP8 uses less memory than FP16 when available, and the grade shows how comfortably each GPU handles the workload.
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 37.2 GB | F | F | F | F |
| 768×768 | 37.4 GB | F | F | F | F |
| 1024×1024 | 37.7 GB | F | F | F | F |
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 19.0 GB | S | F | F | B |
| 768×768 | 19.2 GB | A | F | F | B |
| 1024×1024 | 19.5 GB | A | F | F | B |
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 14.1 GB | S | D | F | A |
| 768×768 | 14.3 GB | S | D | F | A |
| 1024×1024 | 14.6 GB | S | D | F | A |
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 14.1 GB | S | D | F | A |
| 768×768 | 14.3 GB | S | D | F | A |
| 1024×1024 | 14.6 GB | S | D | F | A |
GGUF Q4 available
Quantized GGUF format for lower VRAM and smaller downloads
Turbo / LCM distillation
Use distilled scheduler at 4-8 steps for faster iteration
from diffusers import DiffusionPipeline
import torch
pipe = DiffusionPipeline.from_pretrained(
"ideogram-ai/ideogram-4-fp8",
torch_dtype=torch.float16
)
pipe.to("cuda")
image = pipe(
prompt="a neon sign that reads 'OPEN'",
num_inference_steps=28,
guidance_scale=5.0,
height=1024,
width=1024,
).images[0]
image.save("output.png")Get started
Setup instructions for running Ideogram 4 locally
1. Download the model
Get the checkpoint from HuggingFace
2. Place in:
ComfyUI/models/checkpoints/ (or ComfyUI/models/unet/ for GGUF)3. Launch ComfyUI
python main.pyVRAM allocation at 1024×1024 on RTX 4090 24GB (24 GB)
Time per image at 1024×1024, 28 steps, FP16.
Download Ideogram 4 in the precision that matches your GPU. Lower precision usually means less VRAM pressure, while higher precision keeps more quality.
Community LoRAs (turbo, unconditional, style) available on Hugging Face and CivitAI.
Frequently asked questions
Ideogram 4 (9.3B parameters) requires approximately 37.7 GB of VRAM at FP16 precision for standard 1024×1024 image generation. If you want a lighter setup, lower precisions like FP8 can reduce memory use when available.
Yes, Ideogram 4 can fit on some 8GB GPUs at ~5.8 GB at Q4_K_M. Check the VRAM table above for the exact resolution and precision trade-off.
Ideogram 4 is marked for ComfyUI and Diffusers support in our catalog, so those are the runtimes we recommend first for local setup. If your workflow uses another front end, check the model's available formats and workflow notes above before downloading.
Ideogram 4 is too large for the RTX 4090's 24 GB VRAM at FP16. Consider using FP8 precision or a GPU with more VRAM.
There are currently no known ControlNet adapters for Ideogram 4. Check Hugging Face and Civitai for community-contributed adapters.
Community LoRAs (turbo, unconditional, style) available on Hugging Face and CivitAI. The LoRA ecosystem for Ideogram 4 is rated as "moderate". Each LoRA adds roughly 0.3 GB of extra VRAM.
On a reference GPU (RTX 4090 24GB), Ideogram 4 generates a 1024×1024 image in approximately ~9s at FP16 with 28 inference steps. Faster GPUs with higher memory bandwidth will produce images more quickly.
See also