OpenAI has launched a new high-resolution AI image generation model, GPT-Image 1.5, marking a significant leap in its ability to understand text prompts, preserve visual details, and process images faster compared to the previous version.
The Most Powerful AI Image Generator Yet
The new model generates images at speeds up to four times faster than before. GPT-Image 1.5 is now available to all ChatGPT users, as well as to developers through the API.
Fidji Simo, CEO of Applications at OpenAI, explained that the new image-generation capabilities are part of a broader transformation of ChatGPT—from an interactive text-based tool into a fully integrated generative interface that automatically activates the appropriate components based on what the user wants to accomplish.
Key Features of GPT-Image 1.5
According to OpenAI, GPT-Image 1.5 can make targeted edits without affecting the rest of the image, while maintaining consistent lighting, composition, and facial features. The model supports adding, removing, merging, blending, and repositioning elements within images, making it a true AI-powered image-editing tool.
Key improvements include:
4× faster image generation compared to the previous model
Higher editing accuracy and better understanding of complex prompts
Preservation of fixed details such as faces, objects, and artistic style
Higher-quality outputs that are cleaner and more detailed
Improved handling of textures, lighting, and fine visual elements
Versatile use cases, including design, marketing, illustrations, UI/UX, and social media
Available free of charge to all ChatGPT users
Greater Adherence to Detailed Instructions
OpenAI also reported that the new model follows detailed instructions more precisely. In a test requiring the arrangement of 36 different elements within a grid, GPT-Image 1.5 successfully positioned the elements accurately—unlike the previous version. This improvement makes it easier to create images that require precise layouts.
The model has also shown enhanced capability in rendering text within images, including smaller and denser text such as article excerpts, short tables, and digital infographics.
Remaining Challenges
Despite these impressive advancements, OpenAI acknowledged that some challenges remain, such as handling very long texts, uncommon fonts, multiple faces within a single image, or generating content across multiple languages.
Overall, GPT-Image 1.5 represents a major step forward in AI-powered image creation and editing, reinforcing ChatGPT’s evolution into a comprehensive generative platform.
Post a Comment