Sora: OpenAI’s New Model for Text-Faithful Image Generation
OpenAI has unveiled a cutting-edge model for image generation that finally delivers on prompt accuracy. It’s a major breakthrough: we can now create visually rich, semantically aligned content with remarkable precision. The long-standing issue of distorted or nonsensical text within images appears to be resolved. You’ll find the new model in Sora under the Images tab.
Sora is an advanced interface built into ChatGPT (available with the Plus and Pro plans) that enables users to generate both images and videos from detailed text prompts. You can also include sketches, reference images, or logos. The system understands context, renders text correctly on whiteboards, signs, and memes, and does so with impressive stylistic coherence.
How do you generate an image?
To create an image, open the side menu, click on “Images,” then switch from “Video” to “Images” next to the prompt input.
From there, simply enter your prompt and wait for the output. Rendering may take a few minutes, as requests are processed in a queue. The wait time depends on overall demand - how many users are accessing Sora at any given moment. For instance, we prompted: “A technology magazine cover with the title ‘AI Revolution: The Future is Now.’” Within seconds, Sora returned an image that aligned perfectly with the request - something that was previously out of reach with DALL·E.
In addition to standard text prompts, Sora allows users to upload a sketch or reference image to guide the generation process. This is especially valuable for designers, marketers, and developers looking to turn rough concepts into production-ready visuals. For example, we uploaded a photo of the Parthenon and asked Sora to convert it into a pencil sketch - the result was outstanding.
Sora opens the door to a wide range of applications. In education and training, it can be used to create instructional infographics that break down complex topics. For example, we asked for a visual explanation of the water cycle, and it produced a clear, well-labeled illustration that captured every stage.
In the marketing world, it’s ideal for generating posters, flyers, social media ads, and product mockups. You can produce a complete advertising visual - featuring a product, a bottle, legible text, and a branded logo - in a single prompt.
Currently, Sora is available exclusively to ChatGPT Plus and Pro users. While usage is unlimited, high traffic may cause delays in generation. Performance tends to improve during off-peak hours, such as late at night in the U.S.
In conclusion, Sora marks a significant shift in how OpenAI approaches visual content generation. It’s a model that not only follows text instructions but also understands sketches, interprets ideas, and transforms them into compelling visuals. Where DALL·E often fell short, Sora excels. It’s no longer just about generating “pretty pictures” - Sora creates purposeful, high-impact content in seconds.
