THUDM/GLM-Z1-Rumination-32B-0414

THUDM/GLM-Z1-Rumination-32B-0414 is a 32 billion parameter model from the GLM-4 family, developed by THUDM, designed for deep reasoning and complex problem-solving. It is specifically trained with "rumination capabilities" to handle open-ended tasks, integrating search tools during its thought process. This model excels in research-style writing, complex retrieval tasks, and function calling, building upon a base pre-trained on 15T high-quality data.