Particle.news
Download on the App Store

Google Rolls Out Gemini 3.1 Pro in Preview With Verified Reasoning Gains

Google is staging the launch to validate agentic workflows after integrating its recent Deep Think upgrades.

Overview

  • Gemini 3.1 Pro is available today in the Gemini app with higher limits for Google AI Pro and Ultra subscribers, and in NotebookLM for those paid tiers.
  • Developers and enterprises can try the model in preview via the Gemini API in AI Studio, Antigravity, Android Studio, Gemini CLI, Vertex AI, and Gemini Enterprise.
  • Google reports a verified 77.1% score on the ARC-AGI-2 logic benchmark, more than doubling Gemini 3 Pro’s reasoning, and 44.4% on Humanity’s Last Exam.
  • The release represents a step forward in core reasoning and remains in preview as Google prepares broader general availability and advances agentic workflows.
  • Google’s demos show practical uses like generating animated SVGs from text and building a live ISS-orbit dashboard from telemetry, while early comparisons note some leaderboards still place rivals slightly ahead.