Sarathi-Serve
AIMade by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made
Sarathi-Serve — chunked prefill, stall-free scheduling
Why it matters
Solved the problem of long prompts stalling decode; 2.6–5.6× the serving capacity of vLLM
Source — Primary sources first; secondary reports are labelled “Press”
Paperarxiv.orghttps://arxiv.org/abs/2403.02310Added in Week 35, 2026 · See that week
On the map — ⑧ Inference Infrastructure line
- Previous stopTensorRT-LLM
- This record · stopSarathi-Serve
- Next stopMooncake