GLM 5.2 and the coming AI margin collapse (part 1)
GLM 5.2 is reportedly the first open-weights model that can compete with proprietary frontier models (Opus, GPT) for agentic tasks, while costing roughly 15-20% of the price. The article argues this signals an imminent collapse of AI inference profit margins.
Background
GLM (General Language Model) is a series of large language models developed by the Chinese AI startup Zhipu AI. An "open weights" model means the trained parameters are publicly released, allowing anyone to download, fine-tune, or run it themselves—unlike closed models such as OpenAI's GPT or Anthropic's Claude, which are only accessible via paid APIs. "Agentic work" refers to AI systems that can autonomously plan and execute multi-step tasks (e.g., booking travel or writing code), which is more demanding than simple Q&A. "Inference margins" are the profits companies earn when customers pay per API call. The core claim here is that open models like GLM 5.2 are matching top-tier closed models in capability while costing a fraction of the price, which would squeeze those profit margins and potentially reshape the AI industry's economics.