RadixAttention

technique

Last discussed Oct 24, 2026

SGLang's technique for automatically reusing shared KV cache prefixes across requests by managing contexts in a radix tree.

Recently discussed in

Why Inference Matters 12x More Than Training | YC Paper Club

Liangsheng Yin · Oct 24, 2026

What creators say

Liangsheng Yin

mentioned

Used for: Reusing shared prefixes for agent and sub-agent workloads Praised: Automatic shared prefix reuse; cache-aware scheduling vs: Prior engines could not reuse prefixes efficiently

"At that time there is no way engine can serve the prefix reusing efficiently. So at that time we brought the RadixAttention, which can automatically reuse the shared prefix and compute with the new suffix."

Why Inference Matters 12x More Than Training | YC Paper Club · ▶ 1:32:28 · Oct 24, 2026

Keep exploring

Inference topic · 58 products Llm serving topic · 3 products

Topics

inference llm serving kv cache

About this data

✓45,000+ mentions extracted from real podcast transcripts
✓Every quote links to the source episode
✓Sponsored mentions are always labeled
✓Critical takes are shown, never filtered out

How Followed works →

Get an email when a new creator recommends it