Gemini-3.0 Pro Agentic Tests (& New KingEval): I TESTED Gemini-3 on AGENTIC TESTS & NEW BENCHMARK!



Visit Augment Code: https://www.augmentcode.com/

In this video, I’ll be breaking down the performance of Gemini 3 Pro on my updated KingBench and KingEval setups. It achieved a perfect score on the initial tests and I’ll show you how it dominates the new Godot and Svelte benchmarks while being significantly cheaper than the competition.


Key Takeaways:

🚀 Gemini 3 Pro scores a perfect 100% on KingBench and leads the new KingEval index with a score of 60.4.
💸 It offers incredible price-to-performance, running all benchmarks for just $2.85—way cheaper than Sonnet.
🎮 Introducing new GD Script (Godot) and Svelte benchmarks where Gemini 3 Pro significantly outperforms Opus and GPT-5.1.
🥇 Gemini 3 Pro breaks the 70% threshold on the Agentic Leaderboard, scoring 71.4 and beating Codebuff.
🛠️ Agentic testing with KiloCode shows success in complex tasks like the OpenCode SVG question and Godot game modding.
📉 While it dominates in logic and planning, the Svelte UI generation still lags slightly behind Sonnet visually.
🔥 Gemini 3 Pro is currently available via API and the Antigravity editor, proving to be a top-tier daily driver.

source