OpenAI has unveiled ChatGPT Images, a fresh image generation and editing experience taking direct aim at Google’s oh-so-oddball Nano Banana. The company’s newest imaging flagship model, GPT-Image-1.5, introduces faster rendering times, improved editing controls, and a dedicated Images tab within ChatGPT to facilitate creation and discovery.
The move demonstrates OpenAI’s ambition to compete not only on model quality but also on workflow. More than raw firepower, this release is about precision editing and an interface that encourages experimentation on the fly—precisely where Google has been drawing attention away from its Gemini image tools.
What’s new in ChatGPT Images: speed, edit fidelity, and UI
OpenAI says GPT-Image-1.5 produces visuals “up to 4x faster,” with sharper edit fidelity. Users are able to make photos from scratch or alter existing creations while preserving finer details—such as the ability to adjust lighting, clean up objects, or reimagine a style without distorting faces, textures, or backgrounds.
In ChatGPT, everything related to generation, editing, and inspiration now conveniently lives under a new Images tab. It acts as a discovery feed but also has examples and prompts to get users started with projects—which is helpful for both professional and hobbyist users who want to browse, remix, and iterate without starting from scratch.
OpenAI also highlights the importance of reliability: higher-quality adherence to user intent and more consistent details across a wide variety of outputs. That’s a pain point for systems generating images from text, because characters or motifs often drift between variant versions. If the improvements stick and are spread to more canvases, it would cut down on the number of retries required to zero in on a look.
How it stacks up against Nano Banana and Gemini tools
Nano Banana—industry jargon for Gemini 2.5 Flash Image—is winning over users through its speed, friendliness, and ubiquity. Google has woven it through Gemini, Google Photos, Search, and Messages, making quick touch-ups and prompt-based edits a part of mainstream behavior.
On day-one signal, GPT-Image-1.5 made noise: it quickly shot up to the top of LMArena’s Text-to-Image leaderboard, surpassing Nano Banana Pro. Benchmarks only tell part of the story, but they are a good indication that OpenAI’s model is competitive on quality and prompt-following on commonly used tests.
Early side-by-sides point to parity on basic edits. The night car photo converted into a shot taking place in daylight, for instance, returned similar results on both setups, and user preferences were limited to color balance and shadow recovery. In other words, we are firmly in the land of taste and workflow rather than slam-dunk wins.
How to try ChatGPT Images on web, apps, and via the API
ChatGPT Images is launching in the ChatGPT app and on the web through a new Images tab. It works the same way as usual: limited use for free users, while paid tiers have more generous limits and faster throughput.
Developers can tap GPT-Image-1.5 through the OpenAI API. Pricing is stated by OpenAI, and the model enables standard build-outs such as batch image generation, automatic product shots, and embeddable editors for creative tools and design pipelines.
Why this launch matters for creators, marketers, and teams
The image race is now compressing around three levers: latency, edit fidelity, and distribution. Google’s bet here is on ubiquity at the intersection of Android and Photos. OpenAI is firing back with faster prompts, finer edits, and a creation feed that’s meant to keep users within ChatGPT—where text, code, and now images share the same roof.
Why creators like it
The real win for creators, marketers, and product teams is time saved; the hope is that if the load test proves out a presumed 4x speedup (which seems to be the case, but we should see), then routine edits and variation generation get materially faster.
The headline is clear: yes, OpenAI now has a Nano Banana rival to be taken seriously, combining strong scores with a creator-friendly flow. GPT-Image-1.5 will become available systemwide for services in the coming weeks, and then a trial period of real-world use—on everything from photo touch-ups to campaign visuals—will test whether GPT-Image-1.5’s instant lead at the milestone translates to everyday preference.
