---
cited_by:
  - '[[thoughts/inference/kimi-k3-prs#4-pr-reading-order|kimik3 §4]]'
date: '2026-11-11'
kind: pr
local: []
merged: '2026-07-29'
number: 50089
source:
  - https://github.com/vllm-project/vllm/blob/7c6729b769597541d327f666273cd972b8a5318d/vllm/models/kimi_k3/nvidia/model.py
  - https://github.com/vllm-project/vllm/blob/7c6729b769597541d327f666273cd972b8a5318d/vllm/models/kimi_k3/nvidia/ops/cute_dsl/latent_moe_tail/fused_add_multicast_gemm.py
  - https://github.com/vllm-project/vllm/blob/7c6729b769597541d327f666273cd972b8a5318d/csrc/libtorch_stable/fused_kimi_k3_mla_key_concat_kv_cache_kernel.cu
  - https://github.com/vllm-project/vllm/blob/7c6729b769597541d327f666273cd972b8a5318d/csrc/libtorch_stable/kimi_k3/fused_kda_decode_kernel.cu
  - https://github.com/vllm-project/vllm/blob/7c6729b769597541d327f666273cd972b8a5318d/vllm/models/kimi_k3/amd/linear.py
  - https://github.com/vllm-project/vllm/blob/7c6729b769597541d327f666273cd972b8a5318d/vllm/models/kimi_k3/nvidia/ops/cute_dsl/latent_moe_tail/allreduce_rmsnorm_reduce_scatter_early_exit.py
  - https://github.com/vllm-project/vllm/blob/7c6729b769597541d327f666273cd972b8a5318d/vllm/models/kimi_k3/nvidia/ops/third_party/kda/chunk.py
  - https://github.com/vllm-project/vllm/blob/7c6729b769597541d327f666273cd972b8a5318d/vllm/models/kimi_k3/amd/ops/third_party/kda/chunk.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#50089 [Model] Add Kimi K3 support: model files and kernels [1/N]'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/50089
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/50089
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/50089.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
Note: it’s not runnable now · full dump: [[thoughts/inference/vllm-prs/50089|#50089]]

