Dhanishtha-2.0-preview is a groundbreaking AI model that introduces 'Intermediate Thinking,' allowing it to pause, reflect, and refine its reasoning multiple times within a single response. This innovative approach enhances the model's efficiency, reducing both time and token usage, and ultimately lowering operational costs. By enabling real-time problem decomposition and iterative refinement, Dhanishtha-2.0-preview excels in complex problem-solving scenarios, making it a valuable tool for tasks requiring deep analysis and self-correction. Its transparent reasoning process, facilitated by dynamic `<think>...</think>` blocks, offers users clear insights into the model's decision-making steps. Additionally, Dhanishtha-2.0-preview supports bilingual reasoning in English and Hindi, with natural code-switching between languages, further broadening its applicability.