Google has announced the launch of Imagen 4, an improved version of the text-to-image model, which, according to the developers, is significantly superior to the previous Imagen 3, especially in the accuracy of rendering text on images.
There are two versions of the model:
* Imagen 4 — optimized for general use, costs $0.04 per image.
* Imagen 4 Ultra — a premium version, $0.06 per image, which demonstrates better text query matching and results, ahead of competitors.
Currently, Imagen 4 is available only through AI Studio, and the Gemini platform does not yet support it, although a release for paid users is expected in the near future. Google is actively investing in generative AI models — not only in text, but also in multimedia. Imagen 4 is the next step in a line that previously did not reach the rendering quality inherent in competitors such as Midjourney or DALL·E. Special attention is now being paid to the accuracy of text rendering in the frame, which has long been a weak point for most generators. However, Google is leaving the model free only temporarily, which can be interpreted as a large beta presentation before the final commercial launch.
The launch of Imagen 4 is a qualitative leap in the competition of generative models, and those who want to appreciate the new level of realism of AI images should hurry up. Free access through AI Studio is a chance to test one of the best models without financial costs. However, in the future we should expect a full transition to a paid basis, which makes the current window of opportunity limited.
36 
45 
51 