Google Unveils Imagen 4 AI Image Generator With Enhanced Text Rendering
Google releases Imagen 4 and Imagen 4 Ultra through the Gemini API, offering improved text rendering and detailed visual outputs. The new models feature SynthID watermarking and compete directly with other leading image generation tools.
Google launches Imagen 4, the latest version of its text-to-image generation technology, making it available through the Gemini API and Google AI Studio. The release features two model variants, including the standard Imagen 4 and the premium Imagen 4 Ultra, which aim to advance the quality and precision of AI-generated visuals. Developers access the standard model at $0.04 per image, while the Ultra version costs $0.06 per image.
The new models provide substantial improvements over previous versions, particularly in rendering textual elements within images. Imagen 4 Ultra demonstrates strong capabilities in handling complex prompts, as shown by sample outputs like multi-panel science-fiction comics, vintage travel postcards, and editorial-style fashion photography. Google plans to introduce additional billing tiers and higher rate limits in the coming weeks.
As part of its responsible AI initiative, Google embeds a non-visible SynthID watermark into all images generated by Imagen 4 to ensure traceability and transparency. This release places Google in direct competition with OpenAI’s DALL·E and Stability AI’s Stable Diffusion, integrating the image generation models into Google's broader generative AI ecosystem.