GPT Image 1.5

• 🆕 OpenAI's text-aware image model, released December 2025.
• 🔤 Crisp typography, infographics, diagrams, and multi-panel layouts.
• 🧠 Built-in reasoning and world knowledge infer scene context.
• 🎯 Improved realism, prompt adherence, and editability versus prior generation.
• 🔧 Region-aware editing preserves composition across multi-step workflows.
• ⚡ Flexible quality–latency tradeoffs for fast or high-fidelity output.
• 🔒 Declines to generate identifiable real people without consent.
• 👁️ Natively multimodal: accepts text and image inputs.

OpenAI

OpenAI is an American artificial intelligence research organization headquartered in San Francisco, structured as both a for-profit public benefit corporation and a nonprofit foundation. The lab developed the GPT family of large language models, the DALL-E…

GPT Image 1.5 is OpenAI's text-to-image generation and editing model, released in December 2025 and exposed through the OpenAI API and ChatGPT. It is a natively multimodal system that accepts both text and reference images, integrating visual and language context to leverage strong world knowledge—for example, inferring that a scene set in Bethel, New York in August 1969 refers to Woodstock without explicit prompting.

Compared with the earlier GPT Image 1, OpenAI describes major improvements in realism, prompt accuracy, and editability. Text rendering is a particular focus: where earlier models treated text as visual patterns, GPT Image 1.5 renders crisp lettering, consistent layouts, and strong contrast, making it well-suited to infographics, diagrams, UI mockups, and marketing materials. The model also adds robust facial and identity preservation for character consistency and region-aware "deterministic" editing, so a single object can be changed while preserving camera angle and lighting.

A companion inpainting variant, GPT Image 1.5 edit, supports image editing through the same API. OpenAI later succeeded this line with GPT Image 2 in 2026; both remain selectable in the image-generation API.

Limitations noted by OpenAI include latency on complex prompts (up to about two minutes) and occasional imprecision in text placement and clarity. Use requires API organization verification, and the model declines to generate identifiable real people without consent.