The latest Data Phoenix news tagged GLM-5.2.
IBM Research and Red Hat say llm-d combined cache-aware routing, distributed KV caching and split inference stages to serve GLM-5.2 on an existing…
The UK AI Security Institute found open-weight models like GLM-5.2 now trail closed frontier models on cyber tasks by four to seven months, down from…
Two independent evaluators found Z.ai's open-weight GLM-5.2 matching Anthropic's Opus 4.8 on cybersecurity tasks at roughly 2.2 times lower API cost.