Adapted by:
Ricardo Aguilar
WriterMobile tech writer and analyst. I studied Psychology, but I've been working in the consumer tech field for the last 10 years. Interested in motor projects and new forms of mobility.
Karen Alfaro
WriterCommunications professional with a decade of experience as a copywriter, proofreader, and editor. As a travel and science journalist, I've collaborated with several print and digital outlets around the world. I'm passionate about culture, music, food, history, and innovative technologies.
Yesterday was Google I/O 2025, and among all the new features the company introduced was Veo 3—its generative AI text-to-video model, which outperformed all competitors by becoming the first to generate audio alongside video.
With the arrival of Veo 3, Veo 2 becomes a more affordable and accessible option. You can try it on iOS and Android via the Gemini app. Here at Xataka On, we put it to the test—and the results are good.
How to access Veo 2. Veo 2 is now part of the Google AI Pro plan, priced at $19.99 a month. This includes access to Gemini 2.5 Pro, Veo 2, Flow, NotebookLM, and Gemini in apps like Gmail and Google Docs, plus 2 TB of cloud storage.
Access to Veo 3 requires the $250 plan, making Veo 2 the more realistic option for users who don’t plan to use this AI professionally.
It’s available on iOS and Android. If you have the Pro plan for Gemini, you’ll see a Video option alongside Research and Canvas when you open the app on Android. Tap it to enter video creation mode with up to eight seconds of generation.
On iOS, go to the top section of the app and switch from the selected model (2.5 Flash, 2.5 Pro, and Deep Research) to Veo 2. Once that’s set, you can enter your prompt.
The instructions. Using Veo 2 in Gemini is easy: Just be as descriptive as possible. The more detailed your input, the better the results.
Testing Veo 2. One of yesterday’s standout meme challenges was a classic test that many AI models still fail: a person eating spaghetti. This absurd benchmark requires complex understanding of textures, motion, and deformation. And yes—Veo 2 pulls it off. Not perfectly, but it gets the job done.
It’s not a perfect AI system. However, the prompt “drone flight over a paradise beach, with people surfing, palm trees, and seagulls flying” produced a jumbled result that lacked coherence. Maybe I should have asked it not to include people—adding too many elements likely distorted the outcome.
In another example, I asked for two people dancing flamenco against a black background. The quality of the hair, makeup, and gesture detail was surprisingly good. However, the framing was so close that the dance itself wasn’t fully visible. Still, it’s another strong demonstration of how this AI operates.
There’s no free bar. I don’t have more examples—not because I didn’t want to try, but because Google didn’t let me. After generating three or four videos, the app said I couldn’t create more for another 30 minutes. Even after waiting, it still wouldn’t let me.
Paying $19.99 doesn’t give you unlimited video generation, so you’ll have to use this tool with some restraint.
Image | Xataka On
Related | Gemini Live Lets Your Phone See—And It’s Both Fascinating and Creepy. Here’s My Experience
View 0 comments