

Google has officially rolled out its Gemini 3 Deep Think artificial intelligence model to eligible users. The company first mentioned it last month during the launch of Gemini 3 Pro. This new version focuses on delivering stronger logic, reasoning, mathematics, and science capabilities to solve complex and niche problems. However, Google has restricted access to only its most expensive Google AI Ultra subscription tier.
The model gained attention because of its impressive benchmark scores. On Humanity’s Last Exam, considered one of the toughest academic-reasoning tests for AI, Deep Think scored 41% without any tool use, the highest recorded so far. It also achieved 45.1% on the extremely difficult ARC-AGI-2 reasoning test, which involves visual puzzles and code execution, a score Google called “unprecedented.” These tests are often used to measure how close a model is to artificial general intelligence (AGI).
According to Google, this advanced performance comes from a new technique called parallel reasoning, where the model explores multiple solution paths at the same time instead of following one long chain of thought. Currently, only users paying ₹24,500 per month for the Ultra plan can access Deep Think through the Gemini website or app by selecting “Deep Think” and choosing Gemini 3 Pro.












Comments (0)
No comments yet
Be the first to comment!