OpenAI has unveiled GPT Image 1.5, a new AI model specialized for image generation and editing. This model is evaluated as optimized for both commercial and general users, improving upon the limitations of existing image generation capabilities while simultaneously increasing processing speed and accuracy.
GPT Image 1.5, released around the same time as Google's recently announced image generation model, "Nano Banana Pro," features significantly improved text representation within images and understanding of subsequent prompts. OpenAI emphasized that this model excels at editing images containing small text and dense information, making it suitable for complex tasks such as infographic creation.
General users can access GPT Image 1.5 via the image generation function within ChatGPT, while developers can access it via an application programming interface (API). OpenAI has reduced the price of this API by 20% compared to previous models, and claims that the model's improved computational efficiency has also resulted in image generation speeds up to four times faster than before. This will also lead to reduced server costs and energy consumption, making it a significant advantage for businesses.
The new GPT Image 1.5 model also demonstrates its strengths in complex, multi-step image editing. For example, it seamlessly performs the difficult task of extracting elements from three different images, compositing them into a single image, and then batch-changing the overall style. Because it accurately determines which elements to change and which to leave unchanged, it's also suitable for commercial applications that require editing brand images or logos without modification.
OpenAI stated that this model may have limitations in generating images that require specific styles or scientific knowledge, but noted that its error rate on related tasks has been significantly reduced compared to previous models. The new model will be accessible through a separate interface within ChatGPT and will also come with personalized prompt recommendations and image filters.
This announcement comes shortly after the release of the GPT-5.2 model, which demonstrated the ability to solve high-school-level science and mathematics problems, breaking records on AI benchmarks. Building on this achievement, OpenAI recently launched FrontierScience, its own dedicated benchmark, which consists of over 700 questions in physics, chemistry, and biology, to assess the scientific applicability of its algorithms.
With the advancement of AI image generation technology accelerating, the release of GPT Image 1.5 demonstrates OpenAI's commitment to strengthening its leadership in image processing. This model, which simultaneously addresses three key objectives: cost reduction, speed enhancement, and improved accuracy, is highly likely to become a core AI tool in various commercial content production environments.
Get real-time news... Go to TokenPost Telegram
Copyright ยฉ TokenPost. Unauthorized reproduction and redistribution prohibited.
