Key Takeaways:
- Google has unveiled Gemini 3, an upgraded multimodal model with improved reasoning and fluid capabilities across voice, text, and images.
- Gemini 3 introduces generative interfaces that allow the model to autonomously choose the best output format.
- The update includes Gemini Agent, which can handle multi-step tasks within the app and connect to services like Google Calendar and Gmail.
Google has launched Gemini 3, an advanced multimodal model with enhanced reasoning and capabilities across various input formats. The new model features generative interfaces that autonomously determine the best output format and Gemini Agent for handling multi-step tasks within the app. This update also integrates Gemini more deeply into Google's products, offering enhanced features for search, shopping recommendations, and single-prompt software generation.
Insight: Gemini 3 Pro addresses key gaps in earlier models, offering improved visual understanding, code generation, and performance on long tasks, making it appealing for developers of AI applications and agents.
This article was curated by memoment.jp from the feed source: MIT Technology Review.
Read the original article here: https://www.technologyreview.com/2025/11/18/1128065/googles-gemini-3/
© All rights belong to the original publisher.
