Skip to content
CozyGPU

Ideogram 4

Text-to-image with a Turbo / Default / Quality dial and real guidance control, for prompts that have to be followed closely.

Powers
Create Image
Modes
Text to image
GPU class
32 GB Ada Pro or 48 GB Ampere
In a fresh install
Off; one switch in Settings → Models
White lilyThistleOrange tulipPink peonyOrange dahlia
01

What it is

Ideogram 4 is an open-weight text-to-image model with a dual-model guider: unlike the turbo models it keeps classifier-free guidance, so the prompt can be pushed harder when the picture has to match it exactly. A rendering-speed setting trades steps for fidelity: Turbo at 12 steps, Default at 20, Quality at 48.

It sits beside Z-Image Turbo, Qwen Image 2.1 and Krea 2 Turbo in Create Image. A built-in Realism Engine LoRA is there to pick when you want it; the app never applies it on its own.

Model card and weights: Comfy-Org/Ideogram-4
02

Tools it powers

  • Create ImageoptionA picture from a text prompt, with full model settings and LoRAs.Also: Z-Image Turbo, Qwen Image 2.1, Krea 2 Turbo
03

Modes

  • Text to imagePrompt in, image out. Aspect presets; default 1024 x 1024.
04

Settings exposed

  • SizeAspect presets; default 1024 x 1024.
  • Rendering speedTurbo (12 steps), Default (20) or Quality (48).
  • Guidance (CFG)Default 7.
  • SeedRandom, or fixed.
  • LoRAsThe built-in Realism Engine LoRA or one of your own; one per job.
  • Custom LoRAsSupported, one per job. A built-in Realism Engine LoRA ships with the worker; your own .safetensors files load by drag and drop.

Other image models

All models