Essay argues open-weight model economics have collapsed

Category: AI
Status: discussion
Sources: jamesoclaire.com, HN
Summary: A widely discussed post argues that open-weight models from Chinese labs (GLM, DeepSeek, Kimi) now deliver near-frontier coding and agentic quality at roughly an order of magnitude lower cost than proprietary APIs, shifting the economics of building agent products. It is an opinion piece without new benchmarks; it leans on recently published third-party evaluations of GLM-5.2 and similar models.
Why it matters: It frames the cost pressure that open-weight releases are putting on proprietary coding-agent pricing, a recurring theme across recent GLM, Kimi, and Cohere releases.