Google has unveiled a cutting-edge AI model called Gemini 2.0 Flash Thinking, designed to handle complex questions while providing a detailed breakdown of its "thought process," according to a report from . Still in the experimental phase, this advanced reasoning model is expected to rival OpenAI’s o1 reasoning model.

Jeff Dean, Google DeepMind’s chief scientist, shared on X (formerly Twitter) that Gemini 2.0 Flash Thinking is “trained to use thoughts to enhance its reasoning capabilities.” It also leverages the speed of the updated Gemini 2.0 framework. A demo provided by Dean illustrates how the model tackles a physics problem by methodically working through multiple steps before arriving at a solution. While this process isn’t equivalent to human reasoning, it enables the AI to break down complex tasks into smaller, manageable components, leading to more accurate results.

Google product lead Logan Kilpatrick showcased another example where the model successfully reasoned through a problem involving both text and visuals. Kilpatrick noted that this development represents "just the beginning" of Google's advancements in AI reasoning. For those interested, Gemini 2.0 Flash Thinking is available for experimentation on Google’s AI Studio platform.

This announcement follows a series of notable updates in the AI field. Earlier this month, Google introduced the upgraded Gemini 2.0 as part of its broader initiative to develop more “agentic” AI systems. Meanwhile, OpenAI has expanded access to the full version of its o1 reasoning model for ChatGPT subscribers. These innovations mark significant strides in the ongoing evolution of AI reasoning capabilities.