Google just rolled out a massive upgrade to its AI search with Gemini 3. The new model beats ChatGPT 4.0 and GPT-4 Turbo in several benchmark tests. Gemini 3 hits 70 on a tech benchmark compared to 67 and 59 for GPT-4 and GPT-4 Turbo, respectively.
The upgrade powers Google Search and Bard with behind-the-scenes AI muscle. It’s not just better at coding and reasoning; Gemini 3 also scores big on AP Calculus and even demonstrates “emergent creativity.”
The launch follows Google’s push to cement AI leadership after years of playing catch-up. Gemini 3 also claims improved contextual understanding and integrates well with multimodal inputs—images plus text. 
Gemini 3 is built using Google’s Pathways system, a scalable architecture that lets models learn from diverse tasks all at once. This means it can flex across different jobs, from chatbots to creative writing and code generation.
The new tech will roll out across Google’s AI portfolio over the coming months. Google hasn’t announced exact user-facing features yet but promises a smoother, smarter search and chat experience.
Gemini 3 looks like Google’s latest attempt to block OpenAI’s momentum and keep AI search relevant post-ChatGPT hype.
