Recently, Zhipu officially launched its new large model GLM-5.1. With the release of this new product, the model aggregation platform OpenRouter reported that Zhipu's GLM series has increased prices by 10%. After this price adjustment, the token price for core scenarios such as Coding (programming) of GLM-5.1 has been basically aligned with Claude3.5Sonnet from Anthropic. This marks the first time that a domestic large model has achieved price alignment with global top manufacturers in mainstream application scenarios after the early "price war," indicating that the focus of industry competition has shifted from cost dumping to performance premium.
In terms of technical performance, GLM-5.1 set a new world record in the SWE-bench Pro benchmark test, which measures professional software development capabilities, surpassing the previously leading Claude4.6Opus and ranking first among domestic and open-source models. Unlike the current mainstream minute-level interaction logic, GLM-5.1 has strong autonomous engineering capabilities and can work continuously for up to 8 hours in a single task. During long-term operations, the model can independently complete planning, execution, and testing, and has the ability to actively switch strategies and self-repair errors, directly delivering complete engineering-level results.
Due to this news, Zhipu's stock rose as much as 14.12% on the morning of August 8th, closing at 890 yuan per share, with the company's total market value climbing to 396.4 billion yuan. From cutting prices by 90% a year ago to compete for market share, to now leveraging technological advantages to match international benchmarks, Zhipu's strategic transformation reflects how leading domestic manufacturers are attempting to reshape the global AI market value distribution by solving complex tasks with "long-duration" capabilities.
