---
cited_by:
  - '[[thoughts/inference/kimi-k3-prs#4-pr-reading-order|kimik3 §4]]'
date: '2026-11-11'
kind: pr
local:
  - file:///Users/aarnphm/workspace/vllm/vllm/model_executor/models/kimi_linear.py
  - file:///Users/aarnphm/workspace/vllm/vllm/distributed/device_communicators/custom_all_reduce.py
  - file:///Users/aarnphm/workspace/vllm/vllm/model_executor/models/kimi_k25_vit.py
  - file:///Users/aarnphm/workspace/vllm/vllm/v1/attention/backends/mla/rocm_aiter_mla.py
  - file:///Users/aarnphm/workspace/vllm/vllm/model_executor/layers/fused_moe/experts/trtllm_mxfp4_moe.py
  - file:///Users/aarnphm/workspace/vllm/vllm/v1/attention/backends/mla/flashinfer_mla.py
merged: '2026-07-30'
number: 50000
source:
  - https://github.com/vllm-project/vllm/blob/aeeb36b1f17145975c6713242f2447bb8b98782b/vllm/model_executor/layers/mamba/gdn/kimi_gdn_linear_attn.py
  - https://github.com/vllm-project/vllm/blob/aeeb36b1f17145975c6713242f2447bb8b98782b/vllm/model_executor/models/kimi_linear.py
  - https://github.com/vllm-project/vllm/blob/aeeb36b1f17145975c6713242f2447bb8b98782b/vllm/distributed/device_communicators/custom_all_reduce.py
  - https://github.com/vllm-project/vllm/blob/aeeb36b1f17145975c6713242f2447bb8b98782b/vllm/model_executor/models/kimi_k25_vit.py
  - https://github.com/vllm-project/vllm/blob/aeeb36b1f17145975c6713242f2447bb8b98782b/vllm/model_executor/warmup/kimi_k3_triton_warmup.py
  - https://github.com/vllm-project/vllm/blob/aeeb36b1f17145975c6713242f2447bb8b98782b/vllm/v1/attention/backends/mla/rocm_aiter_mla.py
  - https://github.com/vllm-project/vllm/blob/aeeb36b1f17145975c6713242f2447bb8b98782b/vllm/model_executor/layers/fused_moe/experts/trtllm_mxfp4_moe.py
  - https://github.com/vllm-project/vllm/blob/aeeb36b1f17145975c6713242f2447bb8b98782b/vllm/v1/attention/backends/mla/flashinfer_mla.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#50000 [New model] Kimi K3'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/50000
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/50000
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/50000.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
7.30 Note: after merging, it still needs to install to run K3 · full dump: [[thoughts/inference/vllm-prs/50000|#50000]]

