综合性能领先 智谱GLM-4.5登顶HuggingFace Trending榜单

Core Insights - The GLM-4.5 model has half the parameter count of DeepSeek-R1 and one-third of Kimi-K2, yet it outperforms them in multiple benchmark tests due to higher parameter efficiency [3] - The API call pricing for GLM-4.5 is significantly lower than current mainstream models, with input costs at 0.8 yuan per million tokens and output costs at 2 yuan per million tokens [3] - In a performance evaluation covering 12 globally recognized hard tests, GLM-4.5 ranked third globally and first among all domestic and open-source models [3] Model Development - The goal of large language models is to achieve human-level cognition across a wide range of fields rather than being designed for specific tasks [4] - A successful large language model must possess core capabilities such as general problem-solving, generalization, common-sense reasoning, and self-improvement [4] - Current models are not truly general models, as they excel in specific areas like programming or mathematics but do not perform optimally across all tasks [4] Product Availability - The GLM-4.5 model series has been launched on the supercomputing internet AI community, including the base models GLM-4.5 and GLM-4.5-Air, as well as hybrid reasoning models and their FP8 versions [4] - Enterprises and developers can quickly download model files for deployment and fine-tuning from the AI community [4]