Everyone has been going crazy over the upcoming Gemini 3.0 family of models from Google.
Especially Gemini 3.0 Pro — this one is going to be absolutely unbelievable — from what internal testers have been saying.
Vibe coding video games from scratch?
And if it’s already doing this internally, it would be interesting to just see how advanced and high-quality those games are.
As always certain people are going to be skeptical of claims like this.
Some have said it’s just hype or the games will be nothing more than Snake and Pacman in terms of complexity.
But they’re forgetting that… current AI models including Gemini can already create these sort of games.
So even “video games” is a bit vague, don’t you think Gemini 3.0 is going to be able to do much more than that.
Even with current AI tech we’ve already had massive success stories — like Pieter Levels who vibe coded a flight simulator in 3 hours — and went on to make over $1 million in a year.
People are already making playable 3d games that generate serious revenue — something like Pacman or Tetris is ALREADY a walk in park for every major model.
People have been making Tetris games with a single prompt with Claude and the rest for months now.
And why do some people think that the only thing Gemini will be able to contribute to the game development is the coding?
Haven’t we already seen the wonders that video generators like Veo and Sora can do?
Is it that hard to imagine progressing from non-interactive video generation to interactive video? AKA, games?
We already have the insane Genie 3.0 from Google — a revolutionary model that generates 3d virtual worlds from scratch — and let you interact with them:
With multimodality and cooperation with Genie, Gemini 3.0 will combine video creation with code generation with audio generation with graphics creation with…
And we’re already getting news that this upcoming model is going to be a significant upgrade in multimodality support.
Even right now people can already create a lot of the static graphics for their game using only these AI models.
We’ve already had major efforts like Blender MCP — that use AI to automate creating high-quality models for 3d models.
So we are clearly already on this trajectory — even Google doesn’t get it right exactly with Gemini 3.0, someone is going to soon enough.
People are still underrating just how quickly these models can progress — even after seeing the unprecedented strides from Midjourney and many others.
We are moving towards fully automated high-quality video game creation.