(Credit: OpenAI)
OpenAI has updated the image-generation tool in ChatGPT with a new flagship model and allocated a dedicated tab for it on the sidebar.
GPT Image 1.5 generates images four times faster than the previous model, OpenAI claims. It's also better at precise edits, meaning it only changes the sections you ask it to adjust, "keeping elements like lighting, composition, and people's appearance consistent across inputs, outputs, and subsequent edits."
OpenAI says the model also excels at adding, subtracting, combining, blending, and transposing. In one example, an image of a person in a group photo is swapped with an anime-style version of the same person, while everyone else remains unchanged.
A new Images tab on the left also opens up a range of preset styling and prompting options. Styling options include sketches, holiday portraits, doodles, and more, while the prompts include "professional job photos," "what would I look like as a K-pop star," and more. The announcement post features multiple styling options applied to OpenAI CEO Sam Altman's face.
A preliminary score from LMArena's text-to-image leaderboard shows GPT Image 1.5 knocking Google's Nano Banana model off the top spot, but that score "may shift as community prompts and votes evolve after public launch," it says.
In a quick test, we asked Image 1.5 to add grass to a lot that already had trees. It added the grass but also changed the shade of the trees' leaves. Nano Banana did not, indicating it may be better at following instructions.
Earlier this month, after Google's Gemini 3 set new industry benchmarks for performance, Altman declared a "code red" and asked employees to focus on improving ChatGPT. OpenAI's latest releases may stem from that urgency. It also rolled out a new GPT 5.2 model last week.
The new image generator is now available for all ChatGPT users, except those on Business and Enterprise plans. It will also be available on apps that use its API, such as Wix, Canva, Figma Weave, and others.