InternThinker is a powerful reasoning AI model developed by the Shanghai Artificial Intelligence Laboratory (Shanghai AI Lab), aiming to explore open, controllable, and reliable General Artificial Intelligence (AGI) through a 'fusion of general and specialized knowledge' approach. The model exhibits long-chain thinking capabilities and can engage in self-reflection and correction during the reasoning process, achieving superior results in various complex reasoning tasks such as mathematics, coding, and reasoning puzzles. Its innovation lies in its metacognitive capabilities, allowing it to autonomously generate high-density intellectual data and gather feedback in large-scale sandbox environments, leading to the independent construction of high-quality thinking chains and significantly enhancing the model's performance in complex task handling.