Zhipu AI Launches GLM-5.2 Harness, Offering 5 Million Free Tokens as Anthropic Rivalry Heats Up
Updated
Updated · South China Morning Post · Jul 2
Zhipu AI Launches GLM-5.2 Harness, Offering 5 Million Free Tokens as Anthropic Rivalry Heats Up
3 articles · Updated · South China Morning Post · Jul 2
Summary
Zhipu AI on Wednesday unveiled a harness for its GLM-5.2 model, giving developers a control system to build autonomous coding assistants and AI agents.
The launch targets a fast-growing battleground in automated coding tools, where Chinese and US AI labs are competing to make large language models execute tasks more independently.
To lure developers, Zhipu raised data quotas for existing subscribers by 50% and offered 5 million free tokens to new ZCode users.
The push comes as Anthropic, its US rival, removed hidden code that tracked Chinese users of Claude Code, sharpening scrutiny around the two companies' competition.
Is China's open-source AI a true innovation engine or a clever workaround for US sanctions that mimics American models?
Is the digital copying of AI models the new frontier of corporate espionage and a major national security threat?
GLM-5.2’s 1M-Token Context and MIT License: Z.ai’s Disruptive Entry Amid U.S.-China AI Tensions
Overview
In mid-June 2026, the U.S. Department of Commerce imposed and then lifted export controls on Anthropic’s advanced AI models, creating a volatile global environment and highlighting the need for government oversight as AI systems grew more powerful. This temporary ban and scrutiny fueled geopolitical tensions and left a gap in the market for high-end AI. Zhipu AI seized this moment to launch GLM-5.2, an open-weight model that quickly became a strong competitor to U.S. models. The launch not only offered an alternative for regions affected by U.S. restrictions but also accelerated global AI competition and diversification.