Google has officially released its highly anticipated Gemini 2.5 Deep Think AI to the public, a model that made headlines for achieving gold medal-level performance at the International Mathematical Olympiad (IMO). This cutting-edge AI, developed by Google DeepMind, showcases advanced reasoning capabilities, particularly in complex problem-solving domains like mathematics and coding.
However, there’s a significant catch: the version now available to users is not the same high-performing model that clinched the IMO gold. Instead, Google has rolled out a faster but lower-performing variant, tailored for broader accessibility while sacrificing some of the raw computational power of the competition model.
This public release, available to Google AI Ultra subscribers at a premium cost of $250 per month, still promises impressive results. It leverages parallel thinking and novel reinforcement learning techniques to tackle intricate queries, outperforming many rival models in benchmarks like Humanity’s Last Exam, where it scored a notable 34.8%.
While the toned-down version may disappoint some enthusiasts expecting the full IMO-winning prowess, Google emphasizes that this iteration prioritizes speed and efficiency for practical use. The company hints at ongoing improvements, suggesting that future updates could bring users closer to the capabilities of the original Deep Think competition model.
For now, subscribers can explore Gemini 2.5 Deep Think’s potential in areas like advanced mathematics, coding, and multi-modal reasoning tasks. Google’s move marks a significant step in democratizing elite AI tools, even if the ultimate power remains under wraps.
As the AI race heats up, Google’s strategy of balancing performance with accessibility could set a new standard. Will this compromise satisfy users, or will demand for the full-strength model push Google to rethink its approach? Only time will tell.
Alfred Lee
