Blogger urges readers to avoid OpenAI and Anthropic for open-weight models

The author argues that open-weight models such as GLM-5.3 are about four months behind OpenAI and Anthropic's proprietary models, and recommends running them privately or through providers like Tinfoil.

The post says ChatGPT and Claude keep user chat sessions accessible to the companies, and that OpenAI and Anthropic control access to less restricted cyber-capable models through trusted-access programs.