Google’s Gemini 3 resides as much as the hype and creating video games in a single shot

Google’s Gemini 3 is lastly right here, and we’re impressed with the outcomes, particularly relating to constructing easy video games.

Gemini 3 Professional is a powerful mannequin, and early benchmarks affirm it.

For instance, it tops the LMArena Leaderboard with a rating of 1501 Elo. It additionally gives PhD-level reasoning with prime scores on Humanity’s Final Examination (37.5% with out the utilization of any instruments) and GPQA Diamond (91.9%).

Actual life outcomes additionally again these numbers.

Pietro Schirano, who created MagicPath, a vibe coding device for designers, says we’re coming into a brand new period with Gemini 3.

In his checks, Gemini 3 Professional efficiently created a 3D LEGO editor in a single shot. This implies a single immediate is sufficient to create easy video games in Gemini 3, which is an enormous deal if you happen to ask me.

I requested Gemini 3 Professional to create a 3D LEGO editor.

In a single shot it nailed the UI, advanced spatial logic, and all of the performance.

We’re coming into a brand new period. pic.twitter.com/Y7OndCB8CK

— Pietro Schirano (@skirano) November 18, 2025

LLMs have been historically dangerous with video games, however Gemini 3 reveals some enhancements in that course.

It’s additionally wonderful at video games.

It recreated the outdated iOS sport known as Ridiculous Fishing from only a textual content immediate, together with sound results and music. pic.twitter.com/XIowqGt4dc

— Pietro Schirano (@skirano) November 18, 2025

This aligns with Google’s claims that Gemini 3 Professional redefines multimodal reasoning with 81% on MMMU-Professional and 87.6% on Video-MMMU benchmarks.

“It additionally scores a state-of-the-art 72.1% on SimpleQA Verified, displaying nice progress on factual accuracy,” Google famous in a weblog publish.

“This implies Gemini 3 Professional is very able to fixing advanced issues throughout an enormous array of subjects like science and arithmetic with a excessive diploma of reliability.”

Gemini 3 is spectacular in my early checks, however adherence stays a problem

I have been utilizing Claude Code for a 12 months now, and it has been a terrific assist with my Flutter/Dart initiatives.

Gemini 3 is a greater mannequin than Claude Sonnet 4.5, however there are some areas the place Claude shines.

To date, no mannequin has come near Claude Code, significantly with adherence, and Gemini 3 isn’t any exception.

One of many areas is adherence.

I personally discovered Claude Code higher for following directions. Likewise, Claude Code can be a greater CLI than Gemini 3 Professional, which supplies it an edge over rivals.

For the whole lot else, Gemini 3 is a more sensible choice, particularly if you happen to’ve been utilizing Gemini 2.5 Professional.

If you happen to use LLMs, I would advocate sticking to Sonnet 4.5 for normal duties and Gemini 3 Professional for advanced queries.