QED
Unlike most AI models that simply generate a final answer, Flash Thinking intends to be a little more transparent and show the thinking steps he takes to arrive at a solution. He was also specifically trained to use his thoughts to improve his reasoning, as explained by Jeff Deanscientific manager at Google DeepMind.
Accessible from this address, the model combines deep reasoning capabilities with a lightweight and fast version of Gemini 2.0 Flash, ensuring answers within seconds, even for complex questions. In addition to textual issues, Flash Thinking is supposed to solve questions involving both text and images. For example, it can analyze an image containing numbers and answer a question based on it within seconds.
Impressive demonstrations
Having relegated his unfortunate attempt at Bard to the depths of his cupboards (the first Gemini version which had magnificently crashed during his demonstration), Google Highlights Impressive Gemini 2.0 Flash Thinking Capabilities. The latter solved a complex physics problem in less than 40 seconds, explaining the steps necessary to reach a solution.
The goal is here multiple : help students understand complex concepts by providing detailed and structured explanations, solve problems requiring advanced logic and rapid analysis, finally analyze and solve situations involving multimodal data.
With Flash Thinking, Google strengthens its position against competitors like OpenAI and Anthropic by detailing its way of thinking -like a human you could say. This innovation could represent a direct response to the advanced capabilities of tools like GPT-4.
Tech
Canada
