Google DeepMind has raised the bar in AI-generated video with the release of Veo 3, its most sophisticated video generation model yet. Announced at Google I/O 2025, the new model delivers cinematic-quality videos at 1080p resolution with unprecedented realism and control.
Veo 3's standout feature is its native audio generation capability, allowing it to create videos complete with ambient sounds, sound effects, and even synchronized character dialogue—a key distinction from competitors like OpenAI's Sora. This integration eliminates the need for separate audio production tools, streamlining the creative workflow.
"Veo 3 excels from text and image prompting to real-world physics and accurate lip syncing," explains Eli Collins, Google DeepMind product vice president. The model demonstrates remarkable improvements in prompt adherence, allowing users to specify detailed cinematographic elements like camera movements, lighting, and scene composition with high precision.
Currently available to Google AI Ultra subscribers in the United States at $249.99 per month, Veo 3 comes integrated with Flow, Google's new AI filmmaking tool designed specifically for creative professionals. The platform enables filmmakers and content creators to produce high-quality video content without traditional production resources.
The technology incorporates several safety measures, including SynthID watermarking on every frame to identify AI-generated content. This addresses growing concerns about potential misuse for creating deepfakes or manipulated media.
As AI video generation tools become increasingly sophisticated, the competition between Google and OpenAI continues to intensify. While Sora offers longer video sequences and strong physics simulation, Veo 3's integrated audio and cinematic controls position it as a compelling option for professional content creators looking for all-in-one video production solutions.