Ideogram 4.0 Open Weights and JSON Control Arrive in ComfyUI
ComfyUI integrated Ideogram 4.0 immediately upon its release, enabling local execution of the 9.3B parameter text-to-image model. Creators can now use structured JSON data to control typography and object placement with higher accuracy than standard natural language prompts.
ComfyUI added native support for Ideogram 4.0, the latest open-weights text-to-image model from Ideogram. This integration allows users to run the 9.3B parameter model locally or on private cloud instances, utilizing a new structured prompting method that treats image descriptions like technical specifications. Unlike previous versions that relied solely on natural language, Ideogram 4.0 in ComfyUI interprets JSON-formatted data to dictate the exact placement of text, objects, and stylistic elements.
What's new
Ideogram 4.0 introduces a fundamental shift in how AI models process user intent by moving toward structured control. The model features a 9.3 billion parameter architecture and was released with open weights, making it a direct competitor to Flux.1 and Stable Diffusion 3.5 Large. Within the ComfyUI environment, the model supports a specific JSON-based prompting schema. This allows an editor to define a 'spec sheet' for an image, including precise coordinates for typography and the spatial relationship between different subjects.
As of February 2025, the ComfyUI implementation includes specialized nodes designed for the Ideogram 4.0 workflow. These nodes handle the heavy VRAM requirements of the 9.3B model while providing the interface for the structured JSON prompts. The model is particularly optimized for graphic design tasks, offering significantly better text rendering and layout consistency than older open-source alternatives.
How it fits your workflow
For graphic designers and storyboard artists, Ideogram 4.0 in ComfyUI provides a level of layout predictability that was previously difficult to achieve. While models like Flux.1 Pro excel at photorealism and general prompt adherence, Ideogram 4.0 focuses on the structural integrity of the composition. This makes it an ideal tool for creating posters, UI mockups, and title cards where text placement cannot be left to chance. Instead of 'rolling the dice' with repetitive text prompts, users can specify the exact string and position within the JSON block.
In a professional production pipeline, ComfyUI users can now bridge the gap between generative art and layout design. Ideogram 4.0 serves as a functional alternative to Flux.1 Dev or Stable Diffusion 3.5 for projects requiring heavy typographic integration. Because it is open-weights, studios can fine-tune the model or run it on internal hardware without the per-image costs associated with closed APIs like DALL-E 3 or Midjourney. This local control is essential for VFX houses and agencies that must maintain strict data privacy for client assets.
What it costs / how to try it
Ideogram 4.0 is available as an open-weights model that can be downloaded and run for free within ComfyUI, provided the user has sufficient GPU hardware. Users need to update their ComfyUI installation to the latest version and download the model weights from Hugging Face to access the new nodes and JSON prompting capabilities.
Read the original announcement on ComfyUI ↗