All Things AI · Season 3 · The New Models · Episode 7
Veo and Omni are Google’s video models, and the practical way an agent reaches them is inside two tools already sitting in a Google account: Google Flow, for generating a clip from a prompt, and Google Vids, for building a finished video with avatars, voiceover, and editing. This is how we make our own videos.
What Veo is
Veo 3.1 is Google DeepMind’s current video model, built around the idea that the picture and the sound come out together — sound effects, ambient noise, and dialogue generated with the video rather than added after. It produces 8-second clips at up to 4K, can turn a reference image into a moving scene, can generate the transition between a start frame and an end frame, and can extend a clip into something longer.
In Google Flow, Veo 3.1 comes in three tiers — Lite, Fast, and Quality — each offering 4, 6, or 8-second clips in either aspect ratio.
What Omni is
Gemini Omni is Google’s newer video model, described as a model that can create from any input — image, text, video, or audio — and turn it into one cohesive output. It edits by conversation while keeping a character consistent across the edit, which is a different capability from generating a fresh clip from scratch.
The current version, Omni 1.1 Flash, added scene extension in 10-second steps up to 40 seconds total, first-and-last-frame control, a faster and cheaper 360p draft mode, and up to 3 seconds of video usable as a style reference. It landed inside Google Vids on September 23, 2026.
Why the two live together
Veo generates a scene from a prompt or a reference image. Omni takes what already exists — your own footage, a still image, a voice track — and turns it into a finished, edited clip while holding the subject steady across the cut. In practice you reach for Veo to make a piece of footage exist, and Omni to make an existing piece of footage into the video you’re actually publishing.
How you get to it
Google Flow (labs.google/flow) is the generation tool: with or without a paid subscription you get 50 Flow credits a day, and a subscription adds a monthly pool on top — 200 extra on AI Plus, 1,000 on AI Pro, and 10,000 or 25,000 on the two AI Ultra tiers. Unused credits don’t roll over, so the daily allowance resets whether you spend it or not.
Google Vids (docs.google.com/videos) is the editing tool: it turns a script and a voice track into a finished clip using four modes — Create, Animate, Edit, and Extend — and it now runs on Omni 1.1 for sharper extensions, 1080p output, and precise duration control. Vids also carries AI avatars, over fifty preset options with more than thirty voices, and a personal avatar you can build from your own face and voice.
How we actually use it
We build a still AI image and pair it with our own OmniVoice audio track, then bring both into Vids using the Edit mode, which reads the last several seconds of the uploaded clip and makes the image speak the audio in sync at 1080p portrait. It’s the same idea as an avatar video, built from a still image and a voice track we already own rather than a live recording.
What Vids adds beyond the avatar
Vids builds a full video project, not just one clip: projects run up to 30 minutes, imported audio or video can run up to 95 minutes or 4 gigabytes, and monthly AI limits scale with the plan — from 10 avatar generations and 200 seconds of clips on the smallest business tier up to 500 avatar generations and 10,000 seconds on the largest Ultra tier, with premium tiers offering unlimited AI voiceovers.
A voiceover can also be generated directly from a script up to 2,500 characters, without recording anything yourself, which pairs with the avatar and clip tools to build an entire video from text alone if that’s the starting point.
What to use it for
Match the tool to the job.
- A short generated scene or background clip — Flow, Veo tier depending on how much polish the shot needs.
- A talking-head style video built from a still image and a voice track — Vids, Edit mode, on Omni.
- Extending or upscaling a clip you already have — Vids, the Extend and 1080p-upscale options that came with Omni 1.1.
- A quick draft to check pacing before spending on a full-resolution render — the 360p draft mode, at a fraction of the credit cost.
What it costs
Flow spends credits per generation: Veo 3.1 Lite costs 10 credits, Fast 20, Quality 100 — with Lite and Fast both cut in half on the Ultra plan. Omni in Flow runs 7 to 15 credits at 720p depending on length, or 4 to 7 credits for the cheaper 360p draft. The daily free allowance of 50 credits covers a handful of shorter clips before a subscription tier is worth considering.
Subscriptions run from AI Plus at $4.99 a month up to AI Ultra starting at $99.99. Vids itself carries no separate price beyond the Google Workspace or Google AI plan you’re already on.
Where this goes wrong
One habit worth watching.
Spending Flow credits on a still image. Credits are metered for video generation; a still image renders free through the normal image tools on your own machine. Check the model selector before you send anything through Flow, because the wrong selection quietly burns the daily allowance on the wrong job.
What to try in the first week
Two exercises.
In Flow, generate one short clip on the free daily allowance and watch the credit cost land against the tier you picked, so the Lite-versus-Quality trade-off is a real number instead of a guess.
In Vids, take a still image and a short voice recording you already have, run it through Edit mode, and watch what the sync actually looks like at 1080p before deciding whether this replaces a step you currently pay for elsewhere.
How it sits next to the others
Against a model you reach only through a developer API, Veo and Omni’s advantage for a working agent is that they already live inside two tools Google account holders have open every day — no separate integration, no separate bill to track, credits shared across a plan you may already carry.
Against a single-purpose avatar tool, Omni inside Vids does more of the job in one place: the avatar, the voiceover, and the editing sit together instead of being three separate stops.
Questions agents actually type
What is the difference between Google Veo and Gemini Omni?
Veo generates a new video scene from a text prompt or reference image, with audio built in. Omni takes existing material — footage, a still image, a voice track — and edits or extends it into a finished clip while keeping the subject consistent across the cut.
How do I use Veo or Omni as a real estate agent?
Through Google Flow for generating clips from a prompt, or Google Vids for building a finished video with avatars, voiceover, and editing. Both are reachable from a normal Google account, with a daily free credit allowance in Flow.
How much does Google Veo cost?
In Google Flow, Veo 3.1 Lite costs 10 credits per generation, Fast 20, and Quality 100, with Lite and Fast halved on the AI Ultra plan. Everyone gets 50 free Flow credits daily; subscriptions add a monthly pool on top.
Can Google Vids make a talking-head video from a still image?
Yes, using the Edit mode: upload a short clip built from a still image paired with a voice track, and Vids’ Omni-powered editing makes the image speak in sync at up to 1080p.
Does Google Vids have AI avatars?
Yes — over fifty preset avatars with more than thirty voices, plus a personal avatar built from your own recorded face and voice, all running on Google’s Veo and Omni models.
How many AI videos can I make in Google Vids per month?
Monthly limits scale with the plan: 10 avatar generations and 200 seconds of clips on the smallest business tier, up to 500 avatar generations and 10,000 seconds of clips on the largest AI Ultra tier, with unlimited AI voiceovers on premium tiers.
Can Google Vids generate a voiceover from a script?
Yes. You can record your own voiceover or generate an AI voiceover directly from a script up to 2,500 characters, without recording anything yourself.
Book a 30-minute strategy call →
Season 1 is Claude Code for Real Estate Agents, filmed and on its own hub. All three seasons: AI for real estate agents.