ChatGPT Images 1.5: This is OpenAI's big leap in images

  • ChatGPT Images launches the GPT Image 1.5 model, up to four times faster and with better instruction tracking.
  • The new tool allows precise edits to uploaded photos, maintaining lighting, composition, and facial features.
  • Notable improvement in text generation within images and in complex scenes with many faces or small details.
  • OpenAI launches its own Images section in ChatGPT, now available to most users via API.

ChatGPT Images

AI-powered image generation has become one of the most visible showcases of the race between tech giants. OpenAI has decided to make a move with a deep update of ChatGPT Images, its integrated visual creation system, in a context where models like Google's Nano Banana Pro were dominating much of the conversation.

With this launch, the company behind ChatGPT wants its tool to move beyond being a simple chat add-on and function as a full-fledged feature. a genuine integrated creative studio, faster, more accurate and with an interface designed from scratch to work with images instead of being limited to text.

New GPT Image 1.5 model: speed and precision as its hallmarks

The heart of the update is GPT Image 1.5OpenAI's new flagship model for images. The company claims it can generate visual content up to four times faster than the previous version, something that in practice is especially noticeable during peak hours and on mobile devices, where before it was not uncommon for the process to be interrupted or to take forever when changing applications.

In addition to performance, the key improvement lies in instruction tracking. The system interprets instructions more accurately. complex prompts and precise spatial relationshipsso that requests such as changing only one object, adjusting the lighting, or modifying a person's clothing no longer cause unexpected changes in the rest of the scene.

OpenAI explains that GPT Image 1.5 has been trained to keep crucial image elements constant, such as facial identity, overall composition, or color paletteeven after several rounds of chained editing. This point is especially relevant for professional use, where visual consistency is not a whim, but a requirement.

Spot and chain editing: change only what matters

One of the areas where previous models fell short was the targeted editing of specific areasChanging a hat, adjusting the lighting, or adding an element to the background could end up remixing the entire scene. The new ChatGPT Images directly addresses this problem.

The model is capable of add, remove, combine, mix and transpose elements within the same image while keeping all other important components stable. In practice, this means being able to request actions such as: changing the color of a shirt, modifying a hat, adjusting a traffic sign, or transforming a truck into a fire truck without distorting the rest of the environment.

Behavior in phone calls has also been reinforced chain editionsUntil now, a third or fourth change would usually cause the model to completely "reinvent" the image. With GPT Image 1.5, the tool much more reliably preserves the style, pose, and scene, so you can iterate on the same base without having to start from scratch with each modification.

Creative transformations: from selfie to movie poster

Beyond its technical precision, OpenAI is pushing ChatGPT Images into distinctly creative territory. The system allows users to upload their own photo and, with a relatively simple prompt, obtain the image in a matter of seconds. credible transformed versionsFrom a 90s advertisement to a scene in Times Square in the middle of winter or a Japanese city with a cyberpunk aesthetic.

The model is also capable of recreating specific artistic styles, such as classic movie posters, anime-style illustrations, or historical-looking compositions, respecting key features of the original person. The idea is that the user can "see" themselves in very different contexts, without losing the feeling that it is the same person.

This approach is reminiscent of what models like Nano Banana already offered, but OpenAI is trying to differentiate itself by betting on more controlled conceptual transformationswhere the system maintains the essence of the base photo while changing clothes, environment, lighting or era with considerable visual coherence.

ChatGPT Images says goodbye to the yellowish style and improves complex scenes

For a long time, it was relatively easy to identify if an image had been created with early versions of ChatGPT: they predominated warm tones, creamy finishes, and a certain yellow undertone that revealed its artificial origin. Internal comparisons shown by OpenAI and independent tests, compared to alternatives such as Bing Image CreatorThat trait seems to have been left behind.

The new model offers a more neutral and varied color spectrumThis makes the images look more like conventional photographs unless the user explicitly requests otherwise in the prompt. This helps the images appear less "branded" and more useful in contexts where realism or integration with existing photographic material is desired.

Improvements have also been made to the representation of scenes with many small elementssuch as crowds or backgrounds rich in detail. The faces in large groups are now more distinct from one another, with more natural poses and expressions, and typical flaws such as handprints, tiny strokes, or strange repetitions are reduced.

ChatGPT Images allows you to insert text within images: jump in posters, infographics and mockups

Generating readable text within an image has historically been one of the Achilles' heels of generative AI. OpenAI claims that GPT Image 1.5 takes a significant step forward in this area, with a much more consistent typography rendering than in previous versions.

The model can handle dense, small blocks of textThis opens the door to creating posters, infographics, newspaper page mockups, or designs with tables and markdown-type formats with a level of readability that, while not perfect, is closer to something usable without intensive retouching.

For those working in marketing, education, e-commerce, or digital content, this improvement means reducing the time spent on correct misshapen letters or incomplete wordsIn contexts where there is a need to produce visual materials with clear messages ready for publication, the fact that the model itself generates reasonably clean text becomes a differentiating factor.

A new user experience: a dedicated Images section in ChatGPT

The update doesn't stop at the model; it also affects how it's used. OpenAI has added a new feature to the ChatGPT sidebar. a specific section called “Images”This applies to both the mobile app and the web version. The goal is to separate the visual experience from traditional chat and make it easier for those who don't want to struggle with complex prompts to navigate.

From this new space, the user finds predefined styles, trend suggestions, and templates For frequent tasks such as creating greetings, restoring old photos, switching between different artistic styles, or generating variations of the same product, this approach lowers the barrier to entry for people without technical experience.

Another practical aspect is that the Images section acts as centralized repository of all the user's visual creations. From there it's easier to review previous versions, repeat a style with new content, or continue editing an already generated image, something especially useful in continuous workflows.

From eye-catching accessory to visual work tool

OpenAI itself acknowledges that, until now, image generation within ChatGPT functioned more like a extra eye-catching within an interface designed for text which serves as a solid visual work environment. With this update, the company aims to make a qualitative leap: moving from "test" images for social media to a tool usable in real-world processes.

The improvement in consistency and iteration has a direct impact on sectors such as design, marketing, e-commerce or brandingCompanies that need to adapt the same creative concept to multiple formats, test variations of a product, or maintain the consistency of logos and corporate elements across hundreds of pieces find a clear advantage in this type of control.

Creative platforms operating in Europe, such as web editors and cloud-based design toolsThey are already integrating these models into their workflows. In this area, OpenAI's commitment to a more comprehensive visual environment can be a good fit for both SMEs looking to accelerate the production of graphic materials and internal communications teams at large corporations.

Availability of ChatGPT Images for users, businesses, and developers

OpenAI has begun rolling out the new ChatGPT Images for most users of the platform, including those with free accountsMany users are already seeing a notification when they open the app inviting them to try the image function, and a new dedicated tab in the side menu to centralize its use.

In the business sector, the company has confirmed that advanced access for Business and Enterprise accounts will be rolled out gradually, with a focus on integrations within professional workflowsFor European organizations already using ChatGPT for internal tasks, this means being able to extend its use from text to graphic material generated under the same credentials.

In parallel, GPT Image 1.5 is available through the OpenAI APIThis allows developers to integrate image generation and editing capabilities into their own applications. The company states that the cost of image input and output is approximately 20% lower than the previous model, a significant advantage for large-scale projects or services operating on tight margins.

Competition with Nano Banana Pro and other visual models

OpenAI's move comes at a time of intense competitive pressure. Google has pushed Nano Banana Pro as one of the leading visual generative models, integrated into its ecosystem of creative tools and linked to his Gemini family, which has boosted its use globally.

This situation has led to the establishment of [unclear] in some competing services. strict limits for free usersFor example, by reducing the number of images that can be generated per day, partly due to high demand. In contrast, OpenAI seems to be betting on a combination of broad reach, greater speed, and a more refined editing environment to retain and attract users.

Meanwhile, other players like xAI with its chatbot Grok or various image specialists are pushing for the visual generation becomes a central front in the battle for user attention. OpenAI's strategy involves consolidating ChatGPT as an "all-in-one application," where search, voice, text, images, and video coexist in a single entry point.

With this new ChatGPT Images, OpenAI takes an important step towards a more mature visual toolA faster and more accurate model, a differentiated interface, and editing capabilities clearly geared towards real-world work, both in personal and professional contexts. It remains to be seen to what extent these improvements will be integrated into the daily lives of users and businesses in Spain and Europe, but the message is clear: the image is no longer just a fun addition to chat, but has become a central component of the ChatGPT ecosystem.

ChatGPT creating images
Related article:
ChatGPT now generates images with GPT-4o: everything you need to know