Ideogram 4.0

Open-weight image generation with multilingual text rendering and precise layout control

About model

Ideogram 4.0 is Ideogram's first open-weight text-to-image model, a 9.3B parameter foundation model trained from scratch with best-in-class multilingual text rendering, explicit bounding box controls for precise object and text placement, color palette controls, and native 2K resolution output. It introduces a structured describe-to-structure-to-recreate training approach — the model first reads scenes as structured data, then learns to rebuild images from that representation. Ideogram 4.0 is the top-ranked open-weight model on Design Arena and achieved a 47.9% first-place win rate in ContraLabs' blind typography evaluation judged by professional designers.

Open-Weight Design Arena

Top-ranked open-weight image model by a commanding margin

Typography Win Rate

47.90%

First-place in ContraLabs blind evaluation by professional designers

Native Resolution

From a 9.3B parameter model trained from scratch not a fine-tune

Model key capabilities

Multilingual Text Rendering: Dense, accurate text rendering across languages — leading professional designers in blind typography evaluation with a 47.9% first-place win rate
Bounding Box Layout Control: Explicit coordinate-based placement of objects and text elements for precise compositional control over complex multi-element designs
Color Palette Controls: Structured color specification alongside bounding box layout for full compositional control from prompt to pixel
Native 2K Output: 9.3B parameter foundation model trained from scratch — not a fine-tune — delivering photorealistic 2K images at the frontier of open-weight design generation

API usage

cURL
Python
Typescript

Endpoint:

ideogram/ideogram-4.0

curl -X POST "https://api.together.xyz/v1/images/generations" \
  -H "Authorization: Bearer $TOGETHER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "ideogram/ideogram-4.0",
    "prompt": "Draw an anime style version of this image.",
    "width": 1024,
    "height": 768,
    "steps": 28,
    "n": 1,
    "response_format": "url",
    "image_url": "https://huggingface.co/datasets/patrickvonplaten/random_img/resolve/main/yosemite.png"
  }'

from together import Together

client = Together()

imageCompletion = client.images.generate(
    model="ideogram/ideogram-4.0",
    width=1024,
    height=768,
    steps=28,
    prompt="Draw an anime style version of this image.",
    image_url="https://huggingface.co/datasets/patrickvonplaten/random_img/resolve/main/yosemite.png",
)

print(imageCompletion.data[0].url)

import Together from "together-ai";

const together = new Together();

async function main() {
  const response = await together.images.create({
    model: "ideogram/ideogram-4.0",
    width: 1024,
    height: 1024,
    steps: 28,
    prompt: "Draw an anime style version of this image.",
    image_url: "https://huggingface.co/datasets/patrickvonplaten/random_img/resolve/main/yosemite.png",
  });

  console.log(response.data[0].url);
}

main();

Model card
Architecture Overview:
• 9.3B parameter text-to-image foundation model, trained from scratch — not a fine-tune of any existing model
• Structured JSON prompting interface: describe-to-structure-to-recreate training loop
• Model first reads scenes, backgrounds, text, and objects as structured JSON data, then learns to reconstruct images from that representation
• Supports explicit bounding box layout control for precise object and text placement
• Color palette control for structured color specification
• Native 2K resolution output

Training Methodology:
• Foundation model trained from scratch on diverse visual data spanning photorealism, illustration, typography, and poster design
• Describe-to-structure-to-recreate loop: structured understanding precedes generation, improving layout coherence and text accuracy
• Trained for deep multilingual text rendering across diverse scripts and languages

Performance Characteristics:
• #1 open-weight model on Design Arena by a commanding margin
• 47.9% first-place win rate in ContraLabs blind typography evaluation by 10 professional designers
• Rated 3.55/5 for practical client work usability by professional designers (highest among evaluated models)
• Top-ranked open-weight model on LMArena image leaderboard
‍
Prompting
Together AI API Access:
• Access Ideogram 4.0 via Together AI APIs using the endpoint ideogram/ideogram-4.0
• Authenticate using your Together AI API key in request headers
• Supports text prompts with optional structured JSON for bounding box layout and color palette control
• $0.06 per image on Together AI serverless infrastructure
‍
Applications & use cases
Marketing & Brand Design:
• Posters, banners, and social media assets with accurate multilingual text rendering
• Brand-consistent layouts with precise element placement via bounding box control
• Typography-forward designs rated highest for real-world client usability by professional designers

Infographics & Print:
• Dense text infographics with accurate label placement and multilingual support
• Print-ready 2K assets with structured layout control
• Packaging and editorial design requiring precise compositional accuracy

Product & Ecommerce:
• Product photography and ecommerce visuals with text overlays and labels
• Promotional materials with controlled color palettes and brand typography
• Localized assets across languages without separate per-language pipelines

Creative Production:
• Illustrations, concept art, and photorealistic scenes at native 2K resolution
• Multi-element compositions with object placement and color direction
• Open weights for fine-tuning and domain-specific customization
‍

Related models

Model specifications

Model data

Model provider
Ideogram
Type
Image
Main use cases
Image Generation
Deployment
Serverless
Monthly Reserved
Endpoint
ideogram/ideogram-4.0
Parameters
9.3B
Price
$0.06 / image
Input modalities
Text
Output modalities
Image

Released
June 2, 2026
Category
Image

Run in Playground

Quickstart docs

Deploy model

Ideogram 4.0

About model

API usage

Model card

Prompting

Applications & use cases