Quick summary: Google released Gemini 3 Pro — a more visual, immersive AI with fewer errors and less sycophancy — and Antigravity, a multi-agent development IDE. Both are rolling out experimentally now.
Gemini 3 Pro — Key features
- More visual and immersive responses: improved outputs for *text, images and video*.
- Higher accuracy, less flattery: fewer factual errors and reduced sycophantic tone.
- Where to try it: experimental access in the Gemini app, AI Studio, Vertex AI, and via the API.
- Access tiers: *Pro and Ultra* subscribers can enable Gemini 3 Pro in AI Mode; Gemini 3 powers complex queries in AI Overviews.
Antigravity IDE — What it is
- Multi-agent development environment: agents run autonomously in the editor, terminal and browser.
- Supports multiple models: includes Gemini, *Claude Sonnet 4.5, and GPT-OSS*.
- Cross-platform: available for Windows, macOS, and Linux.
For developers and power users: test Gemini 3 Pro in AI Studio or Vertex AI, and explore Antigravity for multi-agent workflows. (Source: Ars Technica)
