Google Releases Its Most Powerful AI Model, Gemini 3—Here's What You Need to Know

Google has unveiled its most advanced AI model to date, Gemini 3 Pro, boasting capabilities that allow it to process text, images, audio, and video simultaneously, with the capacity to handle up to 1 million tokens. This represents a significant leap over its predecessor, Gemini 2.5 Pro, which scored 21.6% on the Humanity's Last Exam and 4.9% on the ARC-AGI-2 benchmark; Gemini 3 Pro scored 37.5% and 31.1%, respectively. The model employs a sparse mixture-of-experts architecture, activating only a fraction of its parameters for each query, thereby increasing efficiency without sacrificing performance. Gemini 3 is currently available for public testing, although it is not included in the consumer version of the Gemini app and is positioned to compete with AI models like OpenAI’s GPT-5. Google emphasizes that while Gemini 3 excels in technical performance, commercial viability remains a challenge against competitors like ChatGPT. The training utilized extensive data while ensuring safety and quality by filtering out inappropriate content. Gemini 3 Pro can generate up to 64,000 tokens of output, but it may still encounter instances of hallucination or delays during operation.

Source 🔗