Google Vids just got a major video generation upgrade with longer clips and realistic audio

If you have spent any time experimenting with AI video generation over the past year, you know the single biggest headache has always been clip length. Most tools spit out tiny, three- to four-second bursts of motion that feel rushed, forcing you to stitch together dozens of choppy fragments just to explain a simple concept.

Around here lately, we have been keeping a close eye on Google Vids as it steadily matures inside Google Workspace. Just earlier this week, Google added the ability to turn personal photos into custom AI avatars, and last month we saw essential timeline upgrades like point-in-time editing make the editor feel far more like a real production suite.

Now, Google is upgrading the core generative engine under the hood. As announced on the official Google Workspace Updates blog, Google Vids is officially deploying Gemini Omni 1.2 Flash, expanding video clip lengths up to 15 seconds while delivering major improvements to visual continuity and vocal realism.

Longer video generations that actually fit natural pacing

The headline upgrade with Omni 1.2 Flash is simple: you can finally let scenes breathe. Instead of being locked into rigid micro-clips, Google Vids now includes an interactive duration slider that lets you customize generated scenes anywhere from 3 to 15 seconds.

That extra headroom completely changes how you build narrative decks and training walkthroughs, leveraging:

• Natural dialogue pacing: Synthetic narrators have enough room to finish an entire thought or product explanation without the background video cutting abruptly mid-sentence.
• Smoother scene transitions: Longer camera pans and steady b-roll shots eliminate the frantic, jump-cut pacing that makes early AI video presentations look amateurish.
• Fewer prompt generations: Generating a single cohesive 15-second scene saves you from writing, re-prompting, and stitching three separate five-second clips together.

Fixing the dreaded character-morphing glitch

Beyond longer runtimes, Gemini Omni 1.2 Flash tackles a classic AI video problem: reference drift. In earlier versions, if you prompted a video featuring a specific person, laptop, or office setting across multiple scenes, their clothing, facial features, or product colors would subtly morph from one cut to the next.

Omni 1.2 Flash brings significantly tighter reference consistency. When you supply image assets or describe a recurring character, the model maintains consistent visual traits across subsequent generations so your spokesperson or product actually looks like the same entity throughout the deck.

Google also overhauled the vocal synthesis pipeline. The new engine provides a wider dynamic range with more natural emotional inflection and vocal cadence, replacing the flat, robotic tone of early synthetic narration with speech that sounds convincingly human.

Google notes that Gemini Omni 1.2 Flash is available immediately in Google Vids across Business Starter, Standard, and Plus tiers, Enterprise accounts, Education Plus, Nonprofits, and Workspace Individual subscribers. Personal Google accounts with Google AI Pro and Ultra subscriptions also have access starting today, with higher generation thresholds reserved for AI Expanded Access add-on licenses.