In this video, I’ll be reviewing the newly launched Gemini 3 model, which is now available on Google’s AI Studio. I’ll cover the pricing, run it through my KingBench 2.0 questions including the floor plan and Blender tests, and compare it against GPT-5.1 and Sonnet to see if it’s truly the new King.
—
Key Takeaways:
🚀 Gemini 3 is finally launched and beats Sonnet and GPT-5.1 on almost every benchmark.
💸 The pricing is incredibly competitive at $2/$12, making it cheaper than Sonnet for the performance.
🎨 The model excels at visual coding, creating perfect SVG Pandas, ThreeJS Pokeballs, and Blender scripts.
🛠️ Google also launched “Antigravity,” a new Agentic IDE that I will be testing separately.
💻 You can use Gemini 3 right now in KiloCode, Gemini CLI, or for free via Zenmux.
💯 KingBench 2.0 is officially retired as Gemini 3 scored a perfect 100% pass rate on the standard tests.
👀 Stay tuned for the upcoming Agentic Benchmarks and the Antigravity deep dive.
source
