Google’s Veo 3 and Imagen 4 Generative AI Models Crank the Realism Dial to 11

May 20, 2025

Jeremy Gray

A collage featuring: colorful salt ponds, a person with a red umbrella in a lavender field, an illustration of raspberries, a close-up of a blue butterfly wing, and a kingfisher diving into water.

At Google I/O 2025 today, the company announced a wide range of new software, technology, and features across its ecosystem, including updates to its generative AI models and tools. Google today introduced Veo 3 and Imagen 4, new versions of its generative video and image models.

Google describes its new models as “significant breakthroughs” in its blog post detailing the announcement.

“These models create breathtaking images, videos and music, empowering artists to bring their creative vision to life. They also power amazing tools for everyone to express themselves,” Google writes.

Google Veo 3

Veo 3, Google’s new “state-of-the-art video generation model,” promises to improve the output quality compared to Veo 2 while also adding audio generation to its videos for the first time. Veo 3 can add relevant background noise and even dialogue between characters.

Google says Veo 3 excels at generating text and photorealistic scenes and is better able to replicate real-world physics than prior Veo models. Google also says Veo 3 can lip sync.

“It’s great at understanding,” Google explains. “You can tell a short story in your prompt, and the model gives you back a clip that brings it to life.”

Veo 3 is available today for Ultra subscribers in the United States through the Gemini app and Flow. It’s also available to enterprise users via Vertex AI. More on Ultra and Flow in a bit.

Google Imagen 4

Google’s latest Imagen model, Imagen 4, promises improved speed and performance. Imagen 4 is better than its predecessor in terms of generating fine details such as intricate fabrics, water droplets, and fur. Google says Imagen 4 is equally suited to photorealistic generations as abstract ones.

A giant whale appears to leap out of the pavement in the middle of an empty city street, surrounded by tall modern buildings, with water splashing around it.

A happy golden retriever with its head out of a car window, ears flapping and tongue out, enjoying a ride on a road with mountains and trees in the background.

A woman with short black hair and bold makeup wears a shimmery, translucent outfit. She stands in a futuristic room with rounded windows, glowing lights, and metallic, crystal-like structures. Misty scenery is visible outside.

A pixel art dinosaur wearing sunglasses relaxes on a sunny beach, builds a sandcastle, jumps over cacti with a crab watching, and finally floats asleep on water, enjoying a vacation. Text bubble: "Even error messages need a vacation.

Imagen 4 can create images in a broader range of aspect ratios and up to 2K resolution. It is also much better at generating text.

Imagen 4 is available today in Gemini, Whisk, and Vertex AI. It is also integrated into Google’s Workspace apps like Docs and Slides. The company will soon release a “fast variant” of Imagen 4 that is up to 10 times faster than Imagen 3.

Google Ultra

As mentioned in the Veo 3 section above, the new video generation model is available now to Google Ultra subscribers. This is a brand new subscription announced today that promises access to Google’s “most capable AI models and premium features.”

Google AI Ultra is available today for a whopping $249.99 per month, although first-time users — which would be everyone right now — can save 50% on the first three months.

Google AI Ultra includes the “best version” of Gemini, including Veo 3, Flow, Whisk, NotebookLM, Gemini across Google’s app ecosystem, Gemini in Chrome, Project Mariner, YouTube Premium, and 30TB of cloud storage for use in Drive, Photos, and Gmail.

Google AI Ultra is only in the U.S. for now, but will come to other countries and regions later.

Google Flow

Veo 3 is also available in Google Flow, which is part of the new Google AI Ultra subscription and the existing Google AI Pro plan. Flow is a new AI filmmaking tool designed for the next generation of creatives.

Google says Flow is “the only AI filmmaking tool custom-designed for Google’s most advanced models — Veo, Imagen, and Gemini.” Flow aims to help storytellers explore ideas and create clips and scenes for their stories.

“Flow is inspired by what it feels like when time slows down and creation is effortless, iterative and full of possibility. It’s custom-designed for Veo, Google’s state-of-the-art generative video model, with exceptional prompt adherence and stunning cinematic outputs that excel at physics and realism. Behind the scenes, Gemini models make prompting intuitive, so you can describe your vision in everyday language. You can bring your own assets to create characters, or use Flow to make your own ingredients with Imagen’s text-to-image capabilities,” Google writes.

Flow includes various tools built for professionals, including direct camera controls, scene extension, asset management, and more. Flow is built upon VideoFX, a Google Labs experiment the company showed off last year.

Google Flow is available today to Google AI Pro and AI Ultra subscribers. It is worth noting that only Ultra users have access to Veo 3 as of now.

Image credits: Google

News, Technology

ai, aiphoto, aivideo, gemini, generativeai, google, googleflow, googleimagen4, googleio2025, googleveo3, imagen4, veo3

Discussion