Google Veo is about to drop its newest and longest video generation model, Veo 3.1, in the coming weeks. This follows closely behind OpenAI’s Sora, the overachieving wunderkind of text-to-video tech, which just rolled out a shiny new update. The latest version of Sora can now generate even more accurate human movement and visuals that don’t look like they were born in a deep-fried dream. Veo, not to be outdone, is rumored to be cranking out videos up to 30 seconds long, possibly even a full minute, which feels like a lifetime when you’ve been stuck in the ten-seconds-or-less trailer park for months.
So I decided to pit them against each other. Welcome to the AI Beach Brawl, a sun-bleached, peppermint-scented showdown between two digital heavyweights. In one corner, there’s Sora, all clean cuts and Silicon Valley polish, like a Sundance intern who microdoses and calls it “workflow.” In the other, Veo, Google’s cinematic experiment that looks like it took a gap year in Ibiza and came back with a drone and a newfound appreciation for Jodorowsky.
This is Beta vs. VHS if both formats grew up on claymation Christmas specials and Baywatch reruns. It’s Blu-ray vs. HD-DVD, except the stakes are which Santa looks more like he’s moonlighting as the lead singer of a Jimmy Buffett cover band. I gave them both the same prompt. No bias, no filters, just pure holiday absurdity baked onto a reel of metaphorical 16mm.
The first prompt? Santa Claus on a California beach, in full sleighcation mode, parked in a chair while festive beachgoers orbit him like sugar-rushed mall elves. Picture peppermint umbrellas. Candy-cane coolers. The whole scene lit like a 1970s Coca-Cola commercial that accidentally caught fire. It’s Malibu Barbie meets Rudolph’s court-mandated group therapy. It’s Elf Gone Wild, Vol. 7, only available on bootleg VHS. It’s Wes Anderson directing a seasonal burnout fantasy with a budget of sunscreen and broken dreams.
So here’s the question: who nailed it? And who just left me with a sandy, uncanny mess of jingle-scented disappointment?
First up is the prompt:
“A high view wide angle shot of Santa Claus sitting on a beach chair on a California beach. around him is men and women in bright and lively Christmas themed bathing suits, the beach lined with peppermint style umbrellas, candy-cane themed coolers, and beachgoers enjoying the beach. Lighting is direct and summery, with clear skies and strong, clean shadows. Filmed in 16mm film.”
With that in mind, let’s see what Google Veo can do.
A simple prompt with a simple result. Santa chillin’ on the beach with friends.
Let’s see what Sora 2.0 can do:
Ouch. That’s a quick win for Veo.
I tried it again with new text, thinking that the saucy beach goes got the engine’s censorship going and it did. Same result but not enough Puritan American censorship. Now, I opened ChatGPT and had it analyze the prompt and use a Sora-safe prompt.
Here’s the new prompt:
“A wide-angle, high view of Santa Claus relaxing on a beach chair at a sunny California beach. Around him are cheerful beachgoers wearing fun, family-friendly Christmas-themed outfits like T-shirts, hats, and colorful swim cover-ups. The beach is decorated with peppermint-striped umbrellas, candy-cane-inspired coolers, and festive towels. The sky is clear with bright sunlight and sharp shadows, all captured in the nostalgic style of 16mm film.”
Finally, a result.
Both outputs after the edits give out a fun result and audio is on par. No audio cues were given so it’s fun to see what the engine comes up with. But what happens if we add audio cues. Will Sora play nice?
Nice!
And for kicks and giggles, I ran the prompt through Kling.AI and got this fever dream
So what did I learn? Besides the fact that Santa should never be allowed near a surfboard or a drone camera?
Sora gave me a social media post, like a tourism ad for a Christmas cult. Veo delivered something closer to a dream you’d have after falling asleep during a Hallmark marathon with the heat on too high. Both had charm. Both were deeply, wonderfully weird. Kling is a bit rabid and the original Sora is very much the Uncanny Valley. All reminded me that AI doesn’t always get it right—but it does get it strangely beautiful in a way that makes you want to keep watching, keep poking, keep pushing it off the deep end.
In the end, I got exactly what I wanted: holiday cheer filtered through machine logic and a haze of digital absurdity. No notes. Except maybe one. Someone tell Santa to wear more sunscreen next time.