What is it:
- We have upgraded Agent Architect to now use Gemini 3 Pro and Flash models to encompass all publicly available levels of reasoning.
Features + Current Display:
- Regarding Agent Architect’s new upgrade, Gemini 3 significantly surpasses Gemini 2.5 with superior complex reasoning, improved factual accuracy (with fewer hallucinations), and enhanced efficiency shown by higher scores on key benchmarks such as Humanity’s Last Exam, MMLU, and AIME.
- In this example, we are provided in natural language and code with an end-to-end guide for building an entire Plugin with all needed supplemental components within Agent Studio:
- The three levels of reasoning (high, medium, and low) comprise of Gemini 3 Pro, Gemini 3 Flash, and Gemini 2.5 Flash Lite, respectively:
