Images · Text → image
Imagen 4
Photorealistic images from Google DeepMind with fine detail and crisp text — in three variants tuned for speed or quality
About the model
Imagen 4 is Google DeepMind’s text-to-image model, introduced at Google I/O 2025. Compared with earlier Imagen versions, it is a clear step forward in photorealism, fine detail, text rendering, and style control.
Our service offers three variants. Imagen 4 Fast is the quickest and most affordable, good for finding an idea and trying different wordings. Imagen 4 is the balanced option for most creative and design tasks. Imagen 4 Ultra delivers maximum detail for final shots, advertising, and product visuals.
The model’s strong suit is in-image text: lettering on posters, packaging, comics, and infographics comes out crisp and legible. Imagen 4 works from a text prompt only, so if you need to build on your own photos or reference images, choose Nano Banana 2 or GPT Image 2.
Strengths
- Photorealism — natural lighting, textures, and fine detail
- Crisp, legible in-image text, including longer lines
- A wide range of styles — from photography to impressionism, surrealism, and abstract art
- Three variants for the job — Fast for exploring ideas, Imagen 4 for balance, Ultra for maximum detail
- Aspect ratios 1:1, 16:9, 9:16, 3:4, 4:3, plus automatic selection
Best for
- Ad visuals and product mockups
- Posters, packaging, comics, and infographics with text
- Landscapes, interiors, and product shots
- Cinematic and editorial visuals
Limitations
- Text prompts only — reference images are not accepted
- Prompt length is limited to 5,000 characters
- Fewer aspect ratios than Nano Banana 2
Prompting tips
- Describe the scene like a photographer — lighting, angle, textures, mood
- Put in-image text in quotes and say where it should appear
- Start with Imagen 4 Fast, then rerun a good prompt in Imagen 4 or Ultra
- Name the style directly — "watercolor", "comic", "studio shot"