OpenAI has rolled out an updated iteration of ChatGPT Images, promising improved instruction-following, more precise editing capabilities, and image generation speeds up to four times faster.
Named GPT-Image-1.5, this new model became accessible on Tuesday to all ChatGPT users and through its API. This release marks the latest development in the intensifying rivalry with Google’s Gemini, following a leaked internal memo last month where OpenAI CEO Sam Altman reportedly declared a “code red.” The memo outlined OpenAI’s strategy to reclaim its top spot in the AI sector after Google started gaining market share with the introduction of Gemini 3, its newest flagship model, and Nano Banana Pro, the most recent version of Google’s popular image generator, both of which have achieved leading positions on the LMArena leaderboard in various benchmarks.
Google continues to hold its dominant position, despite OpenAI’s counter-move last week with the introduction of GPT-5.2, which was presented as its most sophisticated model to date for both developers and general professional applications. While OpenAI had initially scheduled a new image generator release for early January, they accelerated these plans, leading to this week’s announcement. The company’s previous image model, GPT-Image-1, was released in April.
The launch of GPT-Image-1.5 coincides with image and video generation tools evolving past experimental stages to acquire more practical, production-ready functionalities. Similar to Nano Banana Pro, ChatGPT Images now includes post-production features, enabling finer control over edits to ensure visual coherence in elements such as facial features, illumination, arrangement, and color palette throughout various modifications.
This represents a significant advancement, as the majority of generative AI image tools struggle with iterative changes. When prompted for specific adjustments, such as ‘modify the facial expression’ or ‘render the lighting colder,’ these models frequently re-interpret the entire image, resulting in inconsistent outputs.
Beyond new functionalities, ChatGPT Images will now be accessible through a specialized entry point within the ChatGPT sidebar, designed to function “more like a creative studio,” as stated by Fidji Simo, OpenAI’s CEO of applications, in a Tuesday blog post.
Simo elaborated, stating that “The updated image viewing and editing interfaces streamline the process of generating images that align with your creative concept or drawing inspiration from popular prompts and pre-configured filters.”
In addition to the enhanced image generator, OpenAI is rolling out novel methods to enrich the ChatGPT experience through increased visual content. According to Simo, the objective is for search queries to present more visual results accompanied by transparent sources, proving beneficial for activities such as unit conversions or retrieving sports results.
Simo emphasized, “When engaged in creation, you ought to be able to visualize and mold your output. In instances where visuals convey a narrative more effectively than text, ChatGPT should incorporate them. If a swift response is needed or the subsequent action resides in a different application, it should be readily available. By implementing these changes, we can continuously bridge the gap between your conceptual ideas and your capacity to realize them.”