Google Imagen

Imagen started as a research project from Google and evolved through several public releases: Imagen 2 (December 2023), Imagen 3 (August 2024), and the Imagen 4 family (generally available August 2025). The Imagen 4 Fast variant generates images in roughly 2.7 seconds at $0.02 per image, Standard provides 2K resolution at $0.04, and Ultra targets maximum photorealism at $0.06. Google reported 13 billion cumulative images generated through Imagen models by the end of 2025.

Imagen powers image generation in the Gemini app, Google's ImageFX tool, and the Vertex AI (Google's enterprise cloud AI service) and Gemini API for developers. It was designed with strong text rendering and prompt adherence from the start, which positioned it as a serious competitor to Midjourney and DALL-E well before those tools had caught up on those dimensions.

For builders, Imagen is accessible as an API-first tool with clear tiered pricing, making it a natural choice for applications that need to generate images at volume. The photorealistic Ultra tier is particularly used in marketing and advertising workflows. Note that Imagen is distinct from Nano Banana: Imagen 4 is Google's general-purpose text-to-image API line, while Nano Banana refers specifically to the Gemini-native image generation and editing models optimized for conversational, multi-image workflows.