Google has officially launched Imagen 4, the latest version of its text-to-image AI model, alongside a premium variant dubbed Imagen 4 Ultra. These tools promise more accurate image generation from text prompts and are now accessible via the Gemini API and Google AI Studio for paid and limited free usage.
What’s New in Imagen 4 and Imagen 4 Ultra?
The base model, Imagen 4, is described as “your go-to for most tasks” and priced at $0.04 per image. For more complex or specific prompt needs, Imagen 4 Ultra offers higher fidelity outputs at a 50% premium ($0.06 per image).
Google claims both models bring significantly improved text rendering over the previous Imagen 3, making it easier to generate captions, comic text, or visually accurate label details.
Precision Meets Style — But Is It Engaging?
In demos, Imagen 4 Ultra produced visually coherent and prompt-accurate outputs, such as:
- A comic strip featuring a spaceship vs. a giant space lizard, complete with onomatopoeic sound effects like “Crunch!” and “Had!!”
- A vintage travel postcard for Kyoto, precisely showing cherry blossoms, a pagoda, and snow-capped mountains.
- A fake fashion shoot and a hiking couple on a mountain.
While the technical accuracy and detail tracking are notable, some of the images still appear “machine-made” and lack the artistic flair seen in Midjourney 7 or DALL-E 3. Users noted the visual fidelity was good, but the style felt generic or overly polished.
The Changing Perception of AI Art
Despite the upgrades, user excitement around AI-generated images seems to be waning. The broader public sees diminishing novelty in these tools, especially as AI art saturates social feeds and low-quality ad placements.
Even with improved features, Imagen 4 faces stiff competition. While Google’s model does deliver on precision, its outputs may still feel cold compared to the creative edge of rival tools.
