# kvcache-ai/Mooncake

> LLM serving platform built on a KVCache-centric disaggregated architecture: separates prefill and decode and transfers KV cache between nodes over RDMA. Powers Kimi in production at Moonshot AI.

- Magnitude: 1.7 out of 10 — Steady
- Stars: 6,953 total · +7 stars measured 2026-11-02, 00:13–23:10 UTC
- Star trust: star growth looks organic
- Category: Language models · Language: C++ · License: Apache-2.0 · Created: 2024-06-25 · Last push: 2026-11-02
- GitHub: https://github.com/kvcache-ai/Mooncake · Homepage: https://kvcache-ai.github.io/Mooncake/ · Page: https://gitnova.dev/en/r/kvcache-ai/Mooncake

## Useful for

- Deploy PD-disaggregated vLLM or SGLang inference with KV cache transfer via Mooncake
- Speed up weight updates of a large model across thousands of GPUs using P2P Store and RDMA
- Build a distributed KV cache pool shared across inference instances

## Why it’s here

- Star-counter measurements on 2026-11-02 (UTC), 00:13–23:10: 6946 → 6953 stars (+7). This is the change over that interval.

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 1,420
- Issues and pull requests: 5,155
- Watchers: 48
- Average over the last week: 10 per day
- Usual pace: 8 per day
- Stars in the last hour (measured): 0
- Latest release: v0.3.14-pre-4 (2026-11-02)

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-10-04 … 2026-11-02: 8, 5, 5, 10, 10, 7, 12, 8, 10, 9, 13, 10, 7, 6, 8, 10, 10, 10, 10, 5, 4, 6, 10, 14, 9, 13, 6, 9, 12, 7

## Magnitude by day, last 14 days

10-20 1.8, 10-21 1.4, 10-22 1.8, 10-23 1.8, 10-24 1.5, 10-25 1.2, 10-26 1.4, 10-27 2.1, 10-28 2.3, 10-29 2.3, 10-30 2.0, 10-31 1.6, 11-01 1.9, 11-02 1.9

## Similar by description

1. **ai-dynamo/dynamo** — 1.8 · Steady · DevOps & infrastructure · Rust · +10 stars measured 2026-11-02, 00:15–23:10 UTC
   Dynamo is an open-source datacenter-scale inference orchestration layer above SGLang, TensorRT-LLM and vLLM that turns a GPU cluster into a coordinated system with disaggregated serving, KV-aware routing and autoscaling.
   Full card: https://gitnova.dev/en/r/ai-dynamo/dynamo.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-11-02 23:16 UTC, updated every 30 minutes.