ChatGPT Images: OpenAI introduces new image generation model GPT-Image-1.5
OpenAI introduced the new image generation model GPT-Image-1.5. It is said to work faster and more precisely and is likely a response to Google's Nano Banana.
Example of an image generated with OpenAI's new image AI
(Image: OpenAI, Bearbeitung: heise medien)
OpenAI has released a new image generation model called GPT-Image-1.5, which is now available in ChatGPT and via the API. As the company announced on its blog, the model works faster than its predecessor, GPT-Image-1, and delivers more precise results with detailed instructions. Industry experts also see the release as a reaction to Google's image AI Nano Banana Pro, which many considered better than ChatGPT's image generation.
The new model is said to be particularly convincing in preserving lighting, composition, and visual similarity. Image editing has also been improved: elements such as brand logos or central visual components are better preserved during editing, it is stated. OpenAI sees application areas primarily in marketing, for example in the creation of graphics and logos, as well as in e-commerce for product catalogs, variants, and scene shots.
New Image Area in ChatGPT
In parallel with the new model, OpenAI has expanded ChatGPT with a dedicated image area. Users will find a library for saved images, editing interfaces, as well as pre-made filters and prompts that are trending and can serve as a model.
Additionally, ChatGPT now provides visual answers for research, for example, on products or restaurants with photos and specifications. Users can highlight people, places, or products in images to retrieve further information. According to OpenAI, GPT-Image-1.5 generates images up to four times faster than its predecessor.
Videos by heise
App Integrations and Availability
With GPT-Image-1.5, OpenAI is intensifying competition with Google's Gemini, which recently surpassed ChatGPT in several benchmarks. The development aligns with OpenAI's trend towards multimodal models, after GPT-5.2 focusing on office work was already introduced. Commercial use of images created with GPT-Image-1.5 is permitted according to OpenAI, but users are liable for the content. Restrictions apply to the depiction of real people without appropriate rights, as well as for hate content.
(mki)
