At its annual I/O 2025 event, Google introduced major updates to its AI-powered image and video generation technologies, unveiling Imagen 4 and Veo 3, along with a new filmmaking tool called Flow.
These updates bring significant improvements in realism, accuracy, and creative control, as well as audio integration in videos.
In a blog post accompanying the event, Google explained that Imagen 4 builds on the foundation of its predecessor with faster generation speeds and enhanced rendering of text and intricate details. Released nearly a year after Imagen 3, the new model is capable of producing photorealistic and abstract images with more fine-tuned textures such as animal fur, fabric patterns, and water droplets. It also supports a variety of aspect ratios and outputs images in up to 2K resolution.
One of the standout features of Imagen 4 is its advanced handling of text in images. Google noted that it has improved “the spelling of words as well as typography,” and that the model now demonstrates better awareness of “text placement, choice of font size, as well as making creative choices about the font style.”
Currently, Imagen 4 is accessible through the Gemini app, Whisk, Vertex AI for enterprise users, and Google Workspace apps like Docs, Slides, and Vids. It’s yet to be confirmed whether it will roll out to all Gemini users or remain exclusive to paid subscribers. Google also shared plans to launch a future version that generates images 10 times faster than Imagen 3.
Meanwhile, the newly revealed Veo 3 video model takes a leap forward by integrating native audio generation. It can now produce background sounds, ambient noise, and even character dialogue. During a demonstration at the event, Google showcased animated characters engaging in a spoken conversation with “clear and natural-sounding voice.”
Veo 3 also includes advancements in prompt alignment, real-world physics simulation, and lip-sync accuracy. The model is currently available via the Gemini app and the new Flow tool for Google AI Ultra subscribers in the US. Enterprises can also access it through the Vertex AI platform.
Flow, the new AI-assisted video creation app, combines the strengths of Gemini, Imagen, and Veo. It allows users to describe a video scene with natural language, and the app can generate an eight-second clip with consistent visuals—including characters, objects, locations, and styles. Google said Flow has “high prompt adherence” and aims to streamline creative workflows. The app is currently offered to Pro and Ultra plan subscribers in the US.