Zhipu AI Launches First Model Built Exclusively for OpenClaw
Chinese AI company Zhipu AI has launched GLM-5-Turbo, the first general-purpose large language model built exclusively for the OpenClaw agent framework. The model was announced on March 16, 2026, and is specifically optimized for OpenClaw scenarios, strengthening capabilities including tool invo...

Chinese AI company Zhipu AI has launched GLM-5-Turbo, the first general-purpose large language model built exclusively for the OpenClaw agent framework.
The model was announced on March 16, 2026, and is specifically optimized for OpenClaw scenarios, strengthening capabilities including tool invocation, instruction following, and long-chain task execution. According to the company's developer platform, the model was trained specifically for the OpenClaw ecosystem.
Zhipu AI (stock code 02513.HK) saw its shares surge nearly 9% following the announcement, reflecting investor enthusiasm for the OpenClaw-specific optimization. The company also introduced "lobster packages" for individual and enterprise users alongside the new model.
The launch builds on Zhipu's earlier GLM-5 release in February 2026, which was also optimized for working with AI agents including OpenClaw. The new Turbo variant represents a more targeted approach, built exclusively for the OpenClaw workflow.
According to the ZClawBench evaluation benchmark, GLM-5-Turbo shows significant improvements in OpenClaw scenarios compared to its predecessor GLM-5 and outperforms several mainstream competitive models across multiple key task metrics.
The move reflects a broader trend of AI companies optimizing models specifically for agent frameworks, rather than offering general-purpose models. As agentic workflows become more prevalent, model providers are increasingly tailoring their offerings to the specific requirements of autonomous agent systems.

