Veo 3.1: Google releases upgraded AI video generator with more control and realism

Google unveils Veo 3.1, its most advanced AI video generation model

Google has officially launched Veo 3.1, the latest version of its AI-powered video generation platform, featuring enhanced realism, greater creative control, and improved visual quality.
The model is now available through the Gemini API and has also been integrated into Google’s Flow video editor.

Greater control and better prompt accuracy

The Veo 3.1 gives users far more precision and flexibility when creating videos from text prompts.
It can transform static images or short clips into longer, high-quality videos, while maintaining strong visual consistency.

Key improvements include:

• Higher prompt adherence, meaning the AI follows user instructions more accurately;
• Superior image quality and realism in animated or extended clips;
• Support for multiple reference images (or “ingredients”) to define characters, objects, and artistic style;
• Longer video generation, up to one minute, with the ability to extend existing footage;
• Manual control of first and last frames, allowing for seamless transitions;
• Advanced scene editing, including adding or removing specific elements;
• Native synthetic audio generation, automatically syncing sound to every clip.

Smarter visuals and real-world consistency

Google explains that Veo 3.1 automatically refines videos before output, ensuring color, lighting, and physics consistency across frames.
The result is smoother camera motion, realistic shading, and cinematic transitions that make the AI-generated footage look natural and cohesive.

By understanding real-world physics and motion, Veo 3.1 produces videos that feel grounded and visually authentic.

Availability and pricing

The Veo 3.1 model is now available via the Gemini API and the Flow editor.

• Google AI Pro subscribers can access Veo 3.1 Fast, a lighter and faster version for quick results.
• Google AI Ultra users receive full access to the standard Veo 3.1 model with complete functionality.

With Veo 3.1, Google positions itself head-to-head with OpenAI’s Sora 2, intensifying the race for dominance in AI-driven video generation.