GLM-5.2 is a turning point for coding agents. It's the first model a business would actually pay to replace Claude Opus with. We gave GLM-5.2 the ability to control its own context during agentic tasks, and paired it with Subconscious Cache — ensuring reliable long-term memory far beyond the context window.

This blog post walks through our approach to long-horizon agents, long-context LLMs, harness-driven context management, and our model-driven context state compression.

Try GLM-5.2 hosted by Subconscious.dev using the examples in the post. The model keeps running until the message list outgrows your internet connection where your WiFi becomes the bottleneck before context window.