On September 23, 2026, Google began rolling out free AI video generation in Google Vids — powered by Gemini Omni 1.1 Flash — to anyone signed in with a personal Google account, not just paying subscribers. It generates and upscales clips to 1080p, extends scenes, and stamps every output with a SynthID watermark. The free tier is metered, and the "1080p" headline comes with a caveat.
2026-09-28 · by Moe Ameen
On September 23, 2026, Google began rolling out free AI video generation in Google Vids — its Workspace video app at vids.new — to anyone signed in with a personal Google account, not only paying subscribers. The update runs on Gemini Omni 1.1 Flash and was confirmed on Google's official Workspace Updates blog, with the feature becoming visible to accounts over the following one to three days. The pitch is straightforward: high-quality AI video that used to sit behind a Google AI subscription is now something you can try for free by opening vids.new in a desktop browser and choosing "Create AI videos."
What free users get is more than a single generate button. You can produce brand-new AI video scenes and export them in 1080p HD, or take an existing lower-resolution AI clip on your timeline and upscale it to 1080p so it blends with the rest of the edit. Omni 1.1 also brings smoother scene extension — lengthen a clip while keeping the lighting, characters, and environment consistent frame to frame — plus precise duration control so a generated shot lands at an exact length. Every clip is embedded with SynthID, Google DeepMind's invisible AI-provenance watermark.
The free tier is metered rather than unlimited. Google's own announcement doesn't put a number on the free personal-account allowance, but multiple outlets that tested or reported on the rollout put it at around six AI video (clip or avatar) generations a month. Google AI Pro and Ultra subscribers and Workspace Business, Enterprise, Education, and Nonprofit plans get larger generation pools plus admin controls, and an AI Expanded Access add-on raises limits further. One caveat is worth stating plainly about the "1080p" headline: 720p remains the model's default render resolution, 1080p output is reached by generation or upscaling, and some early testers reported clips still coming back at 720p — so treat 1080p as available rather than guaranteed on every render at launch. As always with a phased rollout, confirm the current limits and behavior on Google's own pages before planning work around them.
The useful way to read this news is as an economics shift, not a finished-content shift. Google just made the raw clip free, which means the smart move is to generate liberally in Vids and spend nothing iterating — then hand only the keeper to a tool that finishes and distributes it. That is exactly the seam [Kompozy](/) sits in. Vids gives you a 16:9 export in Drive; Kompozy is the layer that turns it into posts, and it does the work Vids doesn't: it burns in branded, on-style captions (silent-autoplay feeds eat an uncaptioned clip alive), reframes cleanly between vertical and landscape per destination, and layers hook text and lower-thirds through [HyperFrames](/glossary/hyperframes) so the first second reads before anyone taps unmute. Because a SynthID watermark rides along, the [per-post review pipeline](/glossary/autopilot) is where you add an "AI-generated" disclosure to the caption or on-screen text before anything ships — a deliberate step, not an afterthought.
Then it multiplies one clip into a week. From the same idea Kompozy fans out a [Clipped Short](/glossary/clipped-short), a Quote Graphic, a Carousel, native text posts, a blog draft, and a newsletter in your brand voice via a [Persona Brief](/glossary/persona-brief), plus the formats Vids can't make at all — talking-head [Persona Shorts](/glossary/persona-shorts) and HeyGen avatar video. Autopilot schedules and publishes the set across the eight social platforms plus your blog and a Mailchimp newsletter from one queue, so you're not exporting from Drive and re-uploading into six apps by hand. If you want the model detail behind this, see [Gemini Omni 1.1 Flash](/ai-tools/gemini-omni-1-1-flash) and the honest [Google Vids review](/reviews/google-vids); if you're weighing it as your whole content stack rather than a generator, the [Google Vids alternative](/alternatives/google-vids) comparison lays out where each tool stops.
Yes, within a monthly limit. As of September 23, 2026, anyone with a personal Google account can generate AI video in Google Vids at vids.new without a paid subscription, running on Gemini Omni 1.1 Flash. Google's announcement doesn't state a free cap, but reporting on the rollout puts personal accounts at around six generations a month; Google AI Pro/Ultra and Workspace Business/Enterprise plans get larger pools and admin controls.
Google says free users can generate and export in 1080p HD, and can upscale existing lower-resolution AI clips to 1080p. The caveat is that 720p remains the model's default render resolution and some early testers reported clips still coming back at 720p — so 1080p is available at launch rather than guaranteed on every generation. Confirm current behavior on Google's pages.
Yes. Every clip generated in Google Vids carries SynthID, Google DeepMind's invisible, machine-readable watermark that identifies the video as AI-generated. It is not visible to the eye, but you should still treat the output as AI content and add a clear disclosure in the caption or on screen where a platform's rules require it.
No. Google Vids exports a 16:9 video to Google Drive; it does not caption for the feed, reframe to vertical, or publish to any social platform. To turn a Vids clip into captioned, per-platform posts and schedule them across TikTok, Reels, Shorts, X, LinkedIn, and more, you bring the export into a content engine like Kompozy that handles finishing and multi-platform publishing.