SGLang llm-d Support Cuts GLM-5.2 Latency 8.7x
llm-d now treats SGLang as a first-class engine for GLM-5.2. Tiered KV caching cut median time to first token from 174.2 s to 20.1 s in tests.
llm-d now treats SGLang as a first-class engine for GLM-5.2. Tiered KV caching cut median time to first token from 174.2 s to 20.1 s in tests.