OpenAI’s GPT Image 1.5: A Significant Leap Forward in AI Image Generation
OpenAI has unveiled GPT Image 1.5, its latest AI image generation model, marking the company’s third new model release in roughly a month and arriving on the heels of Nano Banana Pro’s impressive launch. Early testing reveals a noticeable improvement over its predecessor, delivering fewer errors, less distortion, and more realistic results – though it still faces stiff competition from Google’s Gemini.
What is GPT Image 1.5 and What Can It Do?
GPT Image 1.5 is now the engine powering image generation within ChatGPT, available to both free and paid users. OpenAI promises a suite of enhancements with this new model, including improved text generation within images, more precise adherence to user instructions, editing capabilities that preserve detail, faster performance, and overall higher-quality outputs with reduced artifacts. Accompanying the model launch is a dedicated “Images” tab within ChatGPT, serving as a central hub for accessing generated images, creating new ones, and exploring various styles and sample prompts.
A Definite Improvement Over Previous Models
Considering OpenAI’s recent model releases have been largely incremental, expectations for GPT Image 1.5 were modest. However, the upgrades are readily apparent. According to one analyst, editing images now yields significantly less distortion, and generating images from scratch feels noticeably faster. The resulting images also exhibit a greater degree of realism.
Generated Images Look More Lifelike
While GPT Image 1.5 represents a clear step forward from OpenAI’s older GPT Image 1 model, the crucial benchmark is its performance against the current leader in AI image creation: Gemini’s Nano Banana Pro. To assess this, a standard test was conducted, prompting each model to generate an image of a “cozy suburban home with an open floor plan, including a living space, dining room, and kitchen.”
GPT Image 1.5’s output demonstrated substantial improvements. The image generated by GPT Image 1 appeared dated, resembling a stock photo and lacking detail. In contrast, the new model produced an image rich in detail and strikingly realistic, comparable to Nano Banana Pro’s result, though minor imperfections – such as conjoined chairs – were still visible.
Further testing involved generating complex illustrations, specifically a six-panel comic depicting a high fantasy world where magic users wield assault rifles against robot knights. GPT Image 1.5 excelled, producing detailed artwork and successfully incorporating a plot twist. However, Nano Banana Pro’s comic remained superior, despite ChatGPT generating seven panels instead of the requested six, resulting in an awkwardly cropped final panel.
Diagram Generation and Editing Capabilities
AI image generators are often challenged by complex diagrams requiring precision and accurate text rendering. When tasked with creating a diagram illustrating a network setup with an Ethernet switch, modem, router, PC, PlayStation, smart light hub, and smart TV, GPT Image 1.5 delivered a more detailed graphic than its predecessor, with no noticeable spelling errors. However, Nano Banana Pro again emerged as the top performer, accurately depicting all connections and maintaining consistent font styling.
Editing capabilities were also put to the test, specifically removing a hand from a nature photograph. GPT Image 1.5 produced an image with superior contrast, vibrant colors, and sharper detail compared to Nano Banana Pro’s attempt. However, the generated area beneath the removed hand didn’t seamlessly blend with the original scene, and the model introduced a slight tessellation effect on finer details while failing to maintain the original image’s aspect ratio.
Similarly, when combining images of chairs and a table with an image of a kitchen, GPT Image 1.5 demonstrated improved distortion control compared to GPT Image 1. While the lighting didn’t perfectly match the existing room, Nano Banana Pro’s result more convincingly integrated the new furniture into the kitchen environment.
GPT Image 1.5 vs. Nano Banana Pro: A Close Race
While GPT Image 1.5 doesn’t consistently outperform Nano Banana Pro, it represents a significant upgrade over its previous iteration. As one source noted, “GPT Image 1.5 will sometimes get things right when Nano Banana Pro stumbles,” highlighting the inherent variability in AI chatbot testing.
A Worthwhile Upgrade for ChatGPT Users
If ChatGPT is your preferred AI chatbot, the new image generation model is a worthwhile upgrade, offering demonstrably better results than before. For those already invested in Gemini, GPT Image 1.5 presents a competitive alternative, particularly as a backup option if free access to Nano Banana Pro is exhausted or if you’re seeking different results from a challenging prompt.
Disclosure: Ziff Davis, PCMag’s parent company, filed a lawsuit against OpenAI in April 2025, alleging copyright infringement in the training and operation of its AI systems.
About Our Expert
Ruben Circelli
Writer, Software
With over a decade of experience writing about consumer technology and video games for publications including Destructoid, GamesRadar+, Lifewire, PCGamesN, Trusted Reviews, and What Hi-Fi?, Ruben Circelli brings a wealth of knowledge to his role as a software reviewer at PCMag. He specializes in AI and productivity software, from chatbots to to-do list apps, and enjoys cooking, gaming, and tinkering with computers in his free time.
