sglang

SKILL.md

High-performance serving framework for LLMs and VLMs with RadixAttention for automatic prefix caching.

What it does: Automatically caches and reuses common prefixes across requests.

| Simple generation | 2500 tok/s | 2800 tok/s | 1.12× | | Few-shot (10 examples) | 500 tok/s | 5000 tok/s | 10× | | Agent (tool calls) | 800 tok/s | 4000 tok/s | 5× | | JSON output | 600 tok/s | 2400 tok/s | 4× |

Fast structured generation and serving for LLMs with RadixAttention prefix caching. Use for JSON/regex outputs, constrained decoding, agentic workflows with tool calls, or when you need 5× faster inference than vLLM with prefix sharing. Powers 300,000+ GPUs at xAI, AMD, NVIDIA, and LinkedIn. Source: orchestra-research/ai-research-skills.

View raw