How Deep Think works: extending Gemini’s parallel “pondering time”
Just as people tackle complex problems by taking the time to explore different angles, weigh potential solutions, and refine a final answer, Deep Think pushes the frontier of pondering capabilities by utilizing parallel pondering techniques. This approach lets Gemini generate many ideas without delay and consider them concurrently, even revising or combining different ideas over time, before arriving at the perfect answer.
Furthermore, by extending the inference time or “pondering time,” we give Gemini more time to explore different hypotheses, and arrive at creative solutions to complex problems.
We’ve also developed novel reinforcement learning techniques that encourage the model to utilize these prolonged reasoning paths, thus enabling Deep Think to develop into a greater, more intuitive problem-solver over time.
