Ideogram has released its 4.0 text-to-image model as open source, featuring 9.3 billion parameters. This new model excels in generating accurate text and complex layouts, achieving a high score on OCR accuracy and ranking second overall in designer preference ELO. It supports structured JSON prompting for precise control over colors, bounding boxes, and text elements, and utilizes a unique single-stream DiT architecture with a Qwen3-VL-8B-Instruct text encoder. AI
IMPACT Accelerates open-source capabilities for complex graphic design and text generation in AI image models.
RANK_REASON Open-source release of a new text-to-image model with detailed technical specifications and benchmark results.
AI-generated summary · Google Gemini · from 5 sources. How we write summaries →
