GLM-5.1 matches Opus agentic performance at 1/3 cost

GLM-5.1 is a category-shifting release that proves open-weights models can now compete with proprietary giants in agentic engineering without the prohibitive price tag. The 66% cost reduction compared to Claude Opus 4.6 makes sophisticated, long-running agents economically viable for small-scale developers and automated production pipelines, while native optimization for 8-hour autonomous execution addresses the "drifting" issues common in traditional LLMs. Achievement of top rankings on SWE-Bench Pro and Uniclaw Arena validates Zhipu AI’s strategy of training on massive domestic chip clusters, and the community's pivot toward environment-driven benchmarks like Uniclaw reflects a growing demand for functional reliability over static leaderboards.