Building on yesterday's ChatGPT image upgrade, OpenAI has detailed GPT Image 1.5, a multimodal model enabling precise conversational photo edits. It responds to rivals like Google's Nano Banana while introducing safeguards against misuse.
OpenAI's image update, rolled out December 16 and detailed further on December 17, introduces GPT Image 1.5—a native multimodal system that treats text prompts and image pixels as unified tokens. This enables seamless conversational edits, such as altering poses, removing objects, adjusting clothing, or refining details while preserving faces, building on the faster generation and instruction-following highlighted previously.
Four times faster and 20% cheaper via API than its predecessor, the model integrates into a new ChatGPT sidebar space with presets and prompts. Fidji Simo, OpenAI's CEO of applications, noted: "Creating and editing images is a different kind of task and deserves a space built for visuals."
The release counters Google's Nano Banana (also called Nano Banana Pro), praised for realistic edits and text rendering since August. GPT Image 1.5 improves in these areas but lags in some drawing styles and scientific accuracy.
Ethical risks loom larger with advanced editing, including deepfakes and non-consensual content. OpenAI deploys filters for sexual/violent material, C2PA metadata (removable), and ongoing refinements. Broader issues include creator backlash over likenesses and copyrights, contrasted by deals like OpenAI's with Disney for 2026 character use amid lawsuits from Ziff Davis.
OpenAI maintains: "We believe we’re still at the beginning of what image generation can enable," signaling more multimodal advancements.
