Machine Learning Street Talk Google Debuts Gemini 3: Mind-Blowing AI Performance
Nov 19, 2025
Gemini 3's launch is a game-changer with its advanced AI reasoning and multi-tasking abilities, surprising early adopters with rapid processing speeds. The comparison with competitors like OpenAI's GPT-5.1 showcases its significant market impact. Notably, it scored 37.4% on the challenging Humanity's Last Exam and excelled in user preference tests. Innovations like Google Antigravity blur the lines between coding and usability, enabling the creation of complex 3D models and simulations. Exciting times lie ahead for AI enthusiasts!
AI Snips
Chapters
Transcript
Episode notes
Rapid Leap In Model Capability
- Gemini 3 marks a significant jump in Google's LLM capabilities just months after Gemini 2.5 and amid rapid frontier releases from competitors.
- The model is being rolled out broadly and positioned as a top contender against GPT-5.1 and Sonnet 4.5.
Massive User Footprint
- Google reports Gemini powering search snippets for over 2 billion monthly users and 650 million monthly app users.
- That scale positions Gemini to quickly challenge ChatGPT's dominant usage metrics.
Video Output Surprise During Testing
- Jaeden tested Gemini 3 Deep Think and had it generate an actual video instead of the requested script, revealing unpredictable multi-modal behavior.
- The result showcased Gemini's ability to produce full videos but highlighted control and prompt specificity issues.