On April 8, Zhipu introduced its latest open-source model, GLM-5.1, which represents the pinnacle of intelligence among its flagship models to date. This model boasts substantial enhancements in long-term task processing abilities, capable of operating independently and continuously for more than eight hours. During this time, it autonomously plans, executes, and evolves, ultimately producing results at a complete engineering level. In benchmark assessments like SWE-Bench Pro, GLM-5.1 delivered exceptional performance, outpacing competitors such as GPT-5.4 and Claude Opus 4.6. Moreover, Zhipu has increased the price of GLM-5.1 by 10%, with the cache hit token price in coding scenarios now nearing that of Anthropic's Claude Sonnet 4.6. This marks the first instance where a domestic large-scale model has achieved price competitiveness with leading international providers in core usage scenarios.