---
cited_by:
  - '[[thoughts/papers/elango2026latentmoeoptimalaccuracyflop#15-latent-moe-plus-situ|kimik3 §1.5]]'
  - '[[thoughts/inference/kimi-k3-prs#4-pr-reading-order|kimik3 §4]]'
date: '2026-11-11'
kind: pr
local: []
merged: '2026-10-08'
number: 56875
source:
  - https://github.com/vllm-project/vllm/blob/8674a794c788d4c7dac10c6f7ddfdae677588d18/vllm/models/kimi_k3/nvidia/model.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#56875 [Perf] Enable Kimi K3 shared expert sharding with DeepEP v2'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/56875
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/56875
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/56875.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
Enable Kimi K3 shared experts sharing with DeepEP V2, previously only enabled for MegaMoE. · full dump: [[thoughts/inference/vllm-prs/56875|#56875]]

