OpenAI rolls out GPT Image 1.5 for in‑chat photorealistic edits

On Tuesday, OpenAI rolled out GPT Image 1.5 to all ChatGPT users, a new image-synthesis model that lets people perform photorealistic image edits and generation by typing prompts.

The model reportedly generates images up to four times faster than its predecessor and costs about 20 percent less through the API. OpenAI had been developing conversational image editing since GPT-4o in 2024, but Google brought a public prototype to market in March and later refined it into the Nano Banana and Nano Banana Pro models. GPT Image 1.5 is a “native multimodal” model that processes text and image pixels inside the same neural network, unlike earlier approaches such as DALL-E 3, which used diffusion methods.

By treating images and text as the same kind of tokens, the model can change a subject’s pose or angle, remove objects, alter visual styles or clothing, and refine specific areas while preserving facial likeness across successive edits. Users can converse with the model about a photograph and iteratively refine edits the way they might workshop an email.

The release represents another step toward making photorealistic image manipulation a casual process that requires no particular visual skills, though the source notes results vary by edit and context. The rollout reached all ChatGPT users on Tuesday; broader usage patterns and downstream impacts are not detailed in the source material.

Key Topics

Tech, Openai, Chatgpt, Google, Nano Banana, Image Synthesis