What Is Veo 4?

Veo 4 is the rumored next-generation AI video model from Google DeepMind, widely expected to debut at Google I/O 2026 on May 19-20. As of May 18, 2026, Google has not published an official announcement, but a steady stream of leaks, UI strings, and reporting points to a major capability jump beyond Veo 3.1.

What Veo 4 is expected to do

Multiple independent reports describe four headline capabilities that separate Veo 4 from the Veo 3.x line:

• Multi-camera scene generation. A single prompt produces footage with multiple camera angles and dynamic switching inside the same clip, replacing the manual stitching required with Veo 3.1.
• Native 4K output. Veo 3.1 supports 4K via upscaling; leaks describe Veo 4 as natively rendering at 4K, with configurable 16:9 and 9:16 aspect ratios.
• Longer clips. Veo 3.1 caps at 8 seconds. Veo 4 is expected to extend to 30-60 second durations with strong temporal consistency.
• Stronger character and audio sync. Veo 3 introduced synchronized audio. Veo 4 is expected to extend this with improved lip sync, better character consistency across scenes, and storyboarding-style controls.

These capabilities are sourced from leaked clips, Gemini app UI strings, and analysis from outlets that cover Google DeepMind closely. They should be treated as expectations, not confirmed specifications, until Google's I/O keynote.

How Veo 4 fits into Google's stack

Google's video AI surface in 2026 has multiple entry points. Veo 4 will likely become the default model in:

• Google AI Studio for direct text-to-video generation
• Vertex AI for enterprise and developer access through APIs
• Google Vids for productivity workflows in Workspace
• Flow, Google's creative editing surface for AI video

There is also a parallel project called Gemini Omni that has appeared in Gemini app UI leaks. Omni is described as a unified multimodal system rather than a specialized video model. The most likely scenario is that Veo 4 powers the high-end specialized video pipeline while Gemini Omni handles cross-modal generation inside the Gemini app. We cover the differences in Veo 4 vs Gemini Omni and What Is Gemini Omni?.

How Veo 4 compares to Happy Oyster

Veo 4 and Happy Oyster sit in different categories. Veo 4 is a video model: it produces 2D clips that play linearly. Happy Oyster, released April 16, 2026 by Alibaba's ATH Innovation Division, is a 3D world simulator: it generates explorable three-dimensional environments with Directing and Wandering modes.

If you need cinematic 4K video, Veo 4 is positioned to be the strongest option once it ships. If you need interactive 3D worlds for games, simulation, or VR, neither Veo 4 nor any other 2D video model fills that need. See Happy Oyster vs Veo 4 for a full feature comparison.

What to watch at I/O 2026

The I/O keynote on May 19, 2026 should clarify several open questions:

1. Naming. Whether the next-gen video model launches as "Veo 4" or under a new Gemini-aligned brand.
2. Pricing. Whether free-tier access continues, and how Veo 4 pricing compares to the current Vertex AI rates ($0.10-$0.60 per second for Veo 3.1).
3. Availability. Whether Veo 4 ships in preview, generally available, or as a research demo on day one.
4. Relationship to Gemini Omni. Whether the two are independent products, layered, or merged.

For ongoing tracking, see Veo 4 release date. For evaluating creative AI tools across video and 3D models, Elser.ai offers cross-platform workflows you can use today.