Google Expands Veo 3.1 with Native Vertical Video Support and 4K Upscaling for Creators

No more hacking 16:9 shots into vertical frames or losing the subject to a clunky center-crop. Google’s Veo 3.1 update ends the era of awkward compositions, introducing native 9:16 support and high-fidelity upscaling to the generative video workflow. This shift transforms Veo from an experimental tool into a professional-grade asset for YouTube Shorts and mobile-first marketing campaigns.

These capabilities are now moving beyond experimental labs and into the broader Google ecosystem, including the Gemini API, Google AI Studio, YouTube, and enterprise environments like Vertex AI and Google Vids.

Native Vertical Video for Social-First Creators

The standout addition is native 9:16 support within the "Veo 3.1 Ingredients to Video" tool. Previously, creators were forced to crop landscape generations, often butchering the frame composition and losing the focal point of the shot. By generating in a native portrait orientation from the start, Veo 3.1 preserves storytelling intent for platforms like TikTok and Instagram.

This "Portrait Mode" goes beyond aspect ratios; it optimizes pixel density and subject placement specifically for vertical viewing. Google is pushing these features directly into the YouTube Shorts and YouTube Create apps, delivering on the promise made last year to put generative video tools into the hands of mobile creators. Users of the Gemini app can also begin experimenting with vertical outputs immediately.

Professional Fidelity: 4K Upscaling and Enhanced 1080p

For editors, this update shatters the 720p ceiling that previously relegated Veo outputs to social experiments. While the base generation remains 720p, new state-of-the-art upscaling techniques push outputs to 1080p and 4K.

Availability for these high-resolution tiers is tiered by platform. Standard Gemini app users remain capped at 720p, while the professional-grade 1080p and 4K upscaling features are reserved for the Gemini API, Vertex AI, and Google’s video editor, Flow. The 1080p tier has also undergone a refinement pass, yielding sharper textures and cleaner edges than earlier versions.

Solving the Consistency Problem

Character "drift" and hallucinated backgrounds have long plagued AI video, making sequential storytelling nearly impossible. The updated "Veo 3.1 Ingredients to Video" tackles this by using reference images—the "ingredients"—to anchor visual consistency.

The model synthesizes these inputs to lock in a character's identity across complex motions or dialogue sequences. This update also allows for the seamless blending of textures and styles, giving developers and creators granular control over the final aesthetic rather than leaving it to the whims of the prompt.

Safety and Technical Specifications

Certain technical guardrails remain in place as the model matures. Generated clips are currently capped at eight seconds, though the Gemini API supports video extensions to build longer narratives. To address transparency and safety concerns, every video produced via Veo 3.1 is embedded with SynthID, Google’s imperceptible digital watermark.

These updates went live yesterday, January 13, 2026, for developers using the Gemini API and Google AI Studio, with a staggered rollout continuing for YouTube and Google Workspace users throughout the week.