SGLang 0.5.21 adds a Decisions API and runtime prefill-decode role switching

The SGLang project published v0.5.21 on GitHub on October 2, 2026. The release notes highlight prefill-decode instances that can switch between prefill and decode without a restart, a Rust core for the prefix cache by default, a new Decisions API (/v1/decisions) that lets an LLM or VLM act as a low-latency classifier and scorer, and a Score API (/v1/score) that scores all candidates in one request. Under breaking changes, request-supplied chat_template values are rejected unless the server starts with --trust-request-chat-template, and deprecated endpoints, environment variables, and aliases, along with legacy radix cache implementations, were removed. The release also adds support for models including DeepSeek-V4.1 Flash and MiMo-V2.6. USASI has not tested these changes.1

USASI. (2026, October 13). SGLang 0.5.21 adds a Decisions API and runtime prefill-decode role switching. United States of America Superintelligence. https://unitedstatesofamericasuperintelligence.com/news/sglang-0-5-21-released/

@misc{usasi_news_sglang_0_5_21_released,
  author = {{USASI}},
  title = {SGLang 0.5.21 adds a Decisions API and runtime prefill-decode role switching},
  year = {2026},
  month = oct,
  howpublished = {\url{https://unitedstatesofamericasuperintelligence.com/news/sglang-0-5-21-released/}},
  note = {United States of America Superintelligence. Last updated 2026-10-13}
}