Desktop Mobile
Artificial Intelligence (AI)

Google Launches Veo 2, Beats Sora in AI Video

Ken Metral

Ken Metral

— 2 min read
Google Launches Veo 2, Beats Sora in AI Video
Credit: Google Deep Mind/Veo 2

Google is stepping up its game in AI video and image generation with the release of Veo 2, the newest version of its video generation model. Positioned to compete directly with OpenAI’s Sora, Veo 2 delivers more realistic, high-resolution videos with improved understanding of real-world physics, human movement, and expression.

Veo 2 is currently available on Google Labs' VideoFX platform but is limited to waitlisted users. Those interested can sign up through a Google Form, with access granted selectively by Google. The model allows creators to explore advanced cinematography techniques:

“Ask it for a genre, specify a lens, suggest cinematic effects and Veo 2 will deliver — at resolutions up to 4K,” Google noted in a blog post.

Competing with OpenAI’s Sora

Veo 2 enters the spotlight following the release of OpenAI's Sora, which has been praised yet criticized for occasional physics inconsistencies and anatomical errors. Google claims Veo 2 outperformed rivals, including Sora, in internal tests measuring “overall preference” and “prompt adherence” by human evaluators.

While Veo 2’s access remains limited, the original Veo is still available via Vertex AI. Videos produced with Veo 2 will include Google’s SynthID watermark to ensure transparency around AI-generated content.

Growing Interest in AI Video Generation

AI video creation has become a key focus for tech leaders and smaller innovators alike:

  • RunwayML launched advanced controls for its Gen-3 Alpha Turbo model.
  • Pika Labs introduced Pika 2.0, giving users more creative control.
  • Luma AI partnered with AWS to bring its Dream Machine to enterprises.

Despite progress, AI video generation still faces challenges, including inconsistencies and public skepticism. Google's Veo 2 aims to address these limitations, though the company admits occasional “hallucinations,” such as extra fingers, remain.

Some filmmakers are beginning to embrace AI’s potential. James Cameron joined Stability AI’s board, and Andy Serkis announced plans for an AI-focused production company.

Imagen 3: Better, Brighter Images

Alongside Veo 2, Google updated its Imagen 3 image generation model. Imagen 3 now produces:

  • More detailed, realistic images
  • Improved prompt accuracy
  • A broader range of art styles, from photorealism to anime

The model is available through ImageFX and Google’s Gemini chatbot on the web.

The Future of AI Creativity

Google's advancements with Veo 2 and Imagen 3 showcase its ambition to lead the generative AI race. With rising adoption among YouTube creators and growing demand for AI-assisted visuals, Google is positioning its models as tools for both creators and innovators.

As competition heats up, Google’s focus on realism, creative control, and user-friendly platforms could set it apart in the AI arena.

Read more